Re: Rotated and scaled CALayer
Hello Andreas, Thank you for detailed explanation. However, I am not able to do what I want. Almost. I have discovered, that if I use only CA layers transformations (via setTransform:), I am getting the same rotated pixels when zooming in. The only way, which does what I want is: a) use core image filter to rotate layer (simple CIAffineTransform) b) use second CI filter to tell renderer to use nearest interpolation instead of linear. c) use CA transformation to scale layer Filter mentioned in b) does nothing. It only sets kCISamplerFilterNearest for kCISamplerFilterMode of the input sampler. And for c) I am using CATransform3DMakeScale. Result is OK - when zooming in pixels remains unrotated. But new problem occurs - when scrolling (even slowly) zoomed in layer, the "pixels" flickers. And that causes filter of b). If I am not using it, o its kCISamplerFilterMode setting to the kCISamplerFilterLinear, flickering is gone. But zoomed pixels becomes rotated. Rotated pixels: http://dl.dropbox.com/u/2030721/ForAppleList/Rotated%20Pixels.png Expected result: http://dl.dropbox.com/u/2030721/ForAppleList/Non-rotated_pixels_when_zooming_in.png Best Regards, Rimas M. p.s. I could provide sample application to demonstrate problem. > - You should add your new layer as a sublayer to the layer of the > view-controller's view - or as a sublayer to any other layer than the > controller view's layer. > > - Usually, you "draw" the contents only once. However, when you set an > *image* via the layers's contents property, you don't need to explicitly draw > the layer - the layer draws the image for you, when it is required. So, just > set the image as the layer's contents. > > - You can scale/zoom the layers's contents by setting the layer's transform > property accordingly. This uses Core Animation, which executes - if possible > - on the GPU, and is therefore the fastest method. Note, this doesn't require > the image to be "drawn" again (it requires "rendering"). > You can create a suitable transform which scales AND zooms at once. > > - A CALayer's properties are "animatable" (see Core Animation). In fact, > changes in properties are animated by default. But you can define any > animation, and CA nicely performs them for you. > > - Using Core Graphics functions is *usually* not as fast as Core Animation > (if a CA alternative exists). But certainly, CG has its use. So, if you want > to scale and zoom interactively, use CA. > > Regarding transforms, see also: > CATransform3DMakeRotation, CATransform3DMakeScale, CATransform3DConcat > > > Regards, > > Andreas ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Updating progress of UIProgressView. And Getting Better Saving Performance
Hello Guys..
I see, I have been using GCD and blocks but on somme OSX app I did before, not
on iPhone, I tough tit wasn't there... :P.
SO, I from within the method( block) that makes all the saving process, I do
something like
>> dispatch_async(dispatch_get_main_queue(), ^{
>> // code executed on main thread goes here (i.e., updating the
>> progress indicator in your case
>>
>> });
and update the progress indicator bar, great im gonna try that..
also I have the option to call from within the same method a [ performSelectorinMainThread: withObject: waitUntilDone:NO];
right?
What about the saving operation improving its saving time?, I will check today
what takes longer, if creating the thumbnial, or encoding the views, maybe I
shouldn't encode the whole view, but s just the image reference and the
transform matrix... ?/ I will try that one also.
Thanks for the replies..
Gustavo
On Dec 7, 2010, at 4:22 AM, davel...@mac.com wrote:
>
> On Dec 6, 2010, at 9:21 PM, Laurent Daudelin wrote:
>
>> On Dec 6, 2010, at 17:16, davel...@mac.com wrote:
>>
>>> On Dec 6, 2010, at 5:37 PM, Gustavo Pizano wrote:
>>>
Hello.
My application is saving some data, and it takes a while to do it, it can
be 1 second to 10 sec around.. Im doing some image processing, The thing
is..
I send the saving operation in another thread using the NSThread +
detachNewThreadSelector:toTarget:withObject: method, and in the main
thread I update a UIActivityIndicator, and stop it when I receive the
NSThreadWillExitNotification. The problem is that when it takes long to
save, it may seem the app is somehow stuck, even the spinning indicator is
running. I wanted to change the ActivityIndicator to a progressview, but
then I can't make it work because the saving process not on the main
thread, i think.. correct me if Im wrong, Im not so much familiar with
multithreaded apps.
As for the saving process, what I do is the following.
I have a Parent view which contains subviews, these subviews are drawing
images. The user can modify this images, (scale and rotate), so when I
save i encode these views so it will save the view's transform, and then
I archive the data I encoded for all these subviews.
>>>
>>>
>>>
>>>
>>> You are correct that you cannot call GUI methods from other threads, but
>>> NSObject (which all your UI objects inherit from) has the method.
>>>
>>> - (void)performSelectorInBackground:(SEL)aSelector withObject:(id)arg
>>>
>>> So from your other thread, you can update the progress indicator by using
>>> it to call a method that updates the progress.
>>>
>>>
>>> This is even easier if you are targeting iOS 4.0 and higher using Blocks
>>> and GrandCentral Dispatch.
>>>
>>> Code typed in email (i.e., not tested):
>>>
>>> dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT,
>>> 0), ^{
>>> // code you want implemented on another thread goes here:
>>>
>>> dispatch_async(dispatch_get_main_queue(), ^{
>>> // code executed on main thread goes here (i.e., updating the
>>> progress indicator in your case
>>>
>>> });
>>> });
>>>
>>> HTH,
>>> Dave
>>
>> Maybe I'm missing something but aren't the UI actions supposed to happen in
>> the main thread, in this case, he should really call
>> "performSelectorOnMainThread:withObject:waitUntilDone:"?
>>
>> -Laurent.
>
>
> I copied and pasted the wrong method. Yes, the onMainThread version is the
> one to use for this.
>
> The Grand Central Dispatch code is ok though as Conrad pointed out.
>
> Dave
>
> ___
>
> Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
>
> Please do not post admin requests or moderator comments to the list.
> Contact the moderators at cocoa-dev-admins(at)lists.apple.com
>
> Help/Unsubscribe/Update your Subscription:
> http://lists.apple.com/mailman/options/cocoa-dev/gustavxcodepicora%40gmail.com
>
> This email sent to gustavxcodepic...@gmail.com
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
On 07/12/2010, at 5:35 PM, Ken Thomases wrote: >> gameColumns = [NSArray arrayWithObjects:allColumns count: boardDimension]; Another issue with this line is that, assuming 'gameColumns' is an instance variable of the object, the array will be autoreleased, leaving the variable pointing at deallocated memory, which will surely crash later. --Graham ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
QT streaming with initWithAttributes
Ok, I have found, in order to use QTMovieDataSizeAttribute I must set QTMovieOpenForPlaybackAttribute to YES. But, when QTMovieOpenForPlaybackAttribute is YES I can't access the Movie variable, as mMovie = (Movie)[mQTMovie quickTimeMovie]; and I need the mMovie variable, e.g. for SetMovieVisualContext(mMovie, visualContext->context); How to deal with this restriction? I am puzzled. Regards -- Leonardo ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
On Dec 7, 2010, at 2:40 AM, Graham Cox wrote: > On 07/12/2010, at 5:35 PM, Ken Thomases wrote: > >>> gameColumns = [NSArray arrayWithObjects:allColumns count: boardDimension]; > > Another issue with this line is that, assuming 'gameColumns' is an instance > variable of the object, the array will be autoreleased, leaving the variable > pointing at deallocated memory, which will surely crash later. Yeah. I got sidetracked by the other issues enough that I didn't even get around to considering the memory management problems. Conrad Shultz addressed another, with the TileColumn instances in the allColumns array. Regards, Ken ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: problem with core data and uno
This is the correct behaviour (at least from Core Data's perspective). Undoing registration effectively amends changes onto the previous change. It doesn't remove the changes entirely from Core Data's notice. On 6 Dec 2010, at 19:40, kvic...@pobox.com wrote: > i'm having a problem with core data and undo. here is the simplified > situation. i've got subclasses of NSManagedObject: object A and object B. > object A has a to-one relationship to object B. the reciprocal relationship > in object B is a to many relationship. i.e., object B can refer to multiple > object A s, but object A can only reference one object B. i then edit object > A (using a sheet to enable editing multiple properties simultaneously) to > refer to a different object B. this is an undoable user action. later on i > programmatically change one of the properties of object B. this is NOT > undoable, and the change is done as follows: > > [moc processPendingChanges]; > [[moc undoManager] disableUndoRegistration]; > > b.propertyB = blah; > ... > > [moc processPendingChanges]; > [[moc undoManager] enableUndoRegistration]; > > now if i subsequently undo the first set of edits to object A, propertyB of > object B gets reverted back to the value it had prior to this entire > sequence. this strikes me as wrong! while the first set of edits (to object > A) did change the to many relationship that object B had, they did not touch > the propertyB property and should not have reverted it. > > it seems to me that core data is saving the entire state of object B, instead > of simply the state of the to many relationship. > > can anyone confirm my suspicions? or suggest that possibly this is a bug in > my app -- i have spent about 12 hours trying to track this down in my app, > and haven't found anything? or if my suspicions are correct, can anyone > suggest a work-around? or is there some other way to do this? > > thanx, > ken > ___ > > Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) > > Please do not post admin requests or moderator comments to the list. > Contact the moderators at cocoa-dev-admins(at)lists.apple.com > > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/cocoa-dev/cocoadev%40mikeabdullah.net > > This email sent to cocoa...@mikeabdullah.net ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Updating progress of UIProgressView. And Getting Better Saving Performance
On Dec 7, 2010, at 3:16 AM, Gustavo Pizano wrote:
> Hello Guys..
>
> I see, I have been using GCD and blocks but on somme OSX app I did before,
> not on iPhone, I tough tit wasn't there... :P.
Blocks/GCD are in iOS 4.0 and higher. Assuming you have a paid developer
account, you may want to watch the relevant WWDC videos from last year.
>
> SO, I from within the method( block) that makes all the saving process, I do
> something like
>
>>> dispatch_async(dispatch_get_main_queue(), ^{
>>>// code executed on main thread goes here (i.e., updating the
>>> progress indicator in your case
>>>
>>> });
>
>
> and update the progress indicator bar, great im gonna try that..
Yes.
>
> also I have the option to call from within the same method a [ With the Progress indicator> performSelectorinMainThread:
> withObject: waitUntilDone:NO];
>
> right?
Yes, ssuming the selector you pass (you left it out) takes the NSNumber and
updates the progress bar.
>
>
> What about the saving operation improving its saving time?, I will check
> today what takes longer, if creating the thumbnial, or encoding the views,
> maybe I shouldn't encode the whole view, but s just the image reference and
> the transform matrix... ?/ I will try that one also.
I didn't look at your code closely and don't have the other code around it to
profile which part is taking so long.
Dave
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
SearchBar doesn't display at all
I have an app that has a TabBar as the rot controller, a NavigationController in the first tab, and a TableView NIB as the first view in the NavigationController (I'll call that the Primary TableView). That TableView provides a list with multiple sections for the user to choose from. When selected, a second NIB with a TableView is pushed onto the screen (I'll call that the Secondary TableView). I only want the SearchBar to show up on the Secondary TableView as that is the real meat of the data. If I add a SearchDisplayController with a UISearchBar to the Primary TableView, it shows up just fine at runtime. However, if I add the SearchDisplayController to the Secondary TableView, it doesn't show up at all at runtime when that view is pushed. Instead, it just shows the TableView. The view looks correct other than that the SearchBar is absent. Any help is appreciated! Jonathan Schmidt ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Problem with NSInputStream
Hi all,
I've a problem with NSInputStream:
In my object I create (only once) an input and output stream like:
CFStreamCreatePairWithSocketToHost(kCFAllocatorDefault,
(CFStringRef)IPString, 2101, &readStream, &writeStream);
if (readStream==NULL || writeStream==NULL ) {
return NO;
}
CFReadStreamSetProperty(readStream,
kCFStreamPropertyShouldCloseNativeSocket, kCFBooleanTrue);
CFWriteStreamSetProperty(writeStream,
kCFStreamPropertyShouldCloseNativeSocket, kCFBooleanTrue);
serialInput=(NSInputStream *)readStream;
serialOutput=(NSOutputStream *)writeStream;
[serialInput setDelegate:self];
[serialOutput setDelegate:self];
[serialInput scheduleInRunLoop:[NSRunLoop currentRunLoop]
forMode:NSDefaultRunLoopMode];
[serialOutput scheduleInRunLoop:[NSRunLoop currentRunLoop]
forMode:NSDefaultRunLoopMode];
[serialOutput open];
[serialInput open];
When I'm done with my object, I clean it up (in dealloc) like:
[serialInput close];
[serialInput release];
[serialOutput close];
[serialOutput release];
The problem is now that the stream event handler is still called after dealloc
has finished, resulting in a crash. How can I make sure that the event handler
is not called anymore? It seems the socket does not get properly closed. Maybe
a problem with slow handshaking over Wi-Fi?
Any help is appreciated.
Kind regards,
Remco Poelstra
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
NSDocument app where the document is a SQLite file
I've written a few small personal Mac apps and one iOS app and now am attempting to write my first NSDocument-based Mac application. I would like the document itself to be a SQLite file (with specific tables for my app - not a generic SQLite file). I've been reading through the NSDocument documentation and it appears this can be done, but I'm not 100% certain of the steps and wanted to make certain I wasn't going down a path I shouldn't before getting too far into it. Here is what I think I need to do: Here's what I think I need to do in my NSDocument subclass. When the user chooses a File->New, the method - (id)initWithType:(NSString *)typeName error:(NSError **)outError is called so in it I would create a SQLite database file in a temporary location and call setFileURL: to store the location of that SQLite file (and store a connection to the SQLite database file). In the - (BOOL)writeToURL:(NSURL *)absoluteURL ofType:(NSString *)typeName error:(NSError **)outError, if the passed absoluteURL matches the return value of fileURL: I would do something like commit the SQLite transactions. If the absoluteURL did not match the returned value of fileURL: I need to copy the current SQLite file to this location. Will this work ok? In the - (BOOL)readFromURL:(NSURL *)absoluteURL ofType:(NSString *)typeName error:(NSError **)outError I would just open the SQLite file and store my connection to it for use in the other methods for updating the file. Am I missing something? Is attempting to use a SQLite file as the document file a bad idea? I'm also considering an iOS application that uses the same SQLite database files so that's why I'd like to use SQLite on the Mac if possible. I'm also not ready to tackle using Core Data although that may be the better option if there are reasons that SQLite doesn't map well to a NSDocument based application. My other option is to not use the NSDocument architecture, and just open one file at a time in my GUI but I thought I'd try to learn about the NSDocument architecture while writing it. Thanks, Dave ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Updating progress of UIProgressView. And Getting Better Saving Performance
David hello.
I didn't use blocks, in order to support 3.2, just in case, you never know who
is out there still using it. so I sue normal [NSObject performSelectorOnThread
methods,..
Now I divided the Saving operation and the thumbnail creation in 2 separate
threads, the thumbnail is inside the thread of saving, now its much faster,
encoding and saving takes little time, I believe that the thumbnail creation is
the responsible for slowing the process.
Basically and I think Im very wrong in whatIm doing is create a screenshot of
the View with all its subviews, following Apple docs to do so, so each view's
layer renderInContext is called. Of course I previously create a
UIImageCOntext, then after I get that image, I scaled it down
to the thumb dimensions, but here is where it get's messy:
The thumb must have a rounded corner mask + the border stroke, if I do that in
the same UIImageContext I created to resize the screenShot I got, then I get no
rounded corners for the image but the stroke of the path is rounded. So I was
reading a little and to add a clip to a context I must have a BItMapContext, in
this case the image gets rounded corners but if I try to stroke the path the
clip I add goes away.
So what I did was, create the thumb with a UIImageContext, then clip it using a
UIBitMapContext, then create a final Image using once again a UIImageContext
which in this case I draw the rounded corner image and then stroke the path.
So I believe I can do better in this process.
Thx
Gustavo
On Dec 7, 2010, at 1:59 PM, David Reed wrote:
>
> On Dec 7, 2010, at 3:16 AM, Gustavo Pizano wrote:
>
>> Hello Guys..
>>
>> I see, I have been using GCD and blocks but on somme OSX app I did before,
>> not on iPhone, I tough tit wasn't there... :P.
>
>
> Blocks/GCD are in iOS 4.0 and higher. Assuming you have a paid developer
> account, you may want to watch the relevant WWDC videos from last year.
>
>>
>> SO, I from within the method( block) that makes all the saving process, I do
>> something like
>>
dispatch_async(dispatch_get_main_queue(), ^{
// code executed on main thread goes here (i.e., updating the
progress indicator in your case
});
>>
>>
>> and update the progress indicator bar, great im gonna try that..
>
>
> Yes.
>
>
>>
>> also I have the option to call from within the same method a
>> [ performSelectorinMainThread:
>> withObject: waitUntilDone:NO];
>>
>> right?
>
>
> Yes, ssuming the selector you pass (you left it out) takes the NSNumber and
> updates the progress bar.
>
>
>>
>>
>> What about the saving operation improving its saving time?, I will check
>> today what takes longer, if creating the thumbnial, or encoding the views,
>> maybe I shouldn't encode the whole view, but s just the image reference and
>> the transform matrix... ?/ I will try that one also.
>
>
> I didn't look at your code closely and don't have the other code around it to
> profile which part is taking so long.
>
> Dave
>
>
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Interprocess Communication (IPC) : Which "technology" to use?
My application needs to communicate with my associated Internet Plugin (a bundle that runs in web browsers). For example, here is one sequence: • App sends a "hello" to the plugin. • Plugin processes "hello", responds to app, indicating that it is installed and alive. • App sends up to a few hundred KB to the plugin. • Plugin sends a few tens of KB back to the app. This may happen several times a day. Apple documentation seems to be steering me into Distributed Objects (DO) for this. But DO are so complicated! There are vended objects, proxies, connection configurations, etc. Looks like I'd have to study it for a week or more. Since the web browser talks in JSON-encoded strings anyhow, I don't need to send any Cocoa objects; just bytes are fine. Trying to use what I know, I've been using NSDistributedNotificationCenter to send little messages, and when there is significant data, the sender writes the data to a temporary file and sends the temporary file's path as userInfo in the notification. Upon receiving the notification, the receiver reads the file; end of story. This seems to work OK, except for the minor annoyance of distributed notifications always arriving on the main thread and needing to be forwarded. I was thinking that maybe I should use NSPort since I've had some experience with it, but NSPort documentation advises that "You should implement interapplication communication using distributed objects whenever possible and use NSPort objects only when necessary." Why is that? How should I be doing this? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Interprocess Communication (IPC) : Which "technology" to use?
On Dec 7, 2010, at 9:28 AM, Jerry Krinock wrote: > My application needs to communicate with my associated Internet Plugin (a > bundle that runs in web browsers). For example, here is one sequence: > > • App sends a "hello" to the plugin. > • Plugin processes "hello", responds to app, indicating that it is installed > and alive. > • App sends up to a few hundred KB to the plugin. > • Plugin sends a few tens of KB back to the app. > > This may happen several times a day. > > Apple documentation seems to be steering me into Distributed Objects (DO) for > this. But DO are so complicated! There are vended objects, proxies, > connection configurations, etc. Looks like I'd have to study it for a week > or more. Since the web browser talks in JSON-encoded strings anyhow, I don't > need to send any Cocoa objects; just bytes are fine. > > Trying to use what I know, I've been using NSDistributedNotificationCenter to > send little messages, and when there is significant data, the sender writes > the data to a temporary file and sends the temporary file's path as userInfo > in the notification. Upon receiving the notification, the receiver reads the > file; end of story. > > This seems to work OK, except for the minor annoyance of distributed > notifications always arriving on the main thread and needing to be forwarded. > I was thinking that maybe I should use NSPort since I've had some experience > with it, but NSPort documentation advises that "You should implement > interapplication communication using distributed objects whenever possible > and use NSPort objects only when necessary." Why is that? > > How should I be doing this? If the data you are exchanging is fairly lightweight I can recommend shared memory segments. Decidedly un-Cocoa-like but very fast and reliable. DO is heavy and distributed notification are unreliable (per the documentation). In both cases you may need code to ensure reliability. Keary Suska Esoteritech, Inc. "Demystifying technology for your home or business" ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Interprocess Communication (IPC) : Which "technology" to use?
Le 7 déc. 2010 à 17:56, Keary Suska a écrit : > On Dec 7, 2010, at 9:28 AM, Jerry Krinock wrote: > >> My application needs to communicate with my associated Internet Plugin (a >> bundle that runs in web browsers). For example, here is one sequence: >> >> • App sends a "hello" to the plugin. >> • Plugin processes "hello", responds to app, indicating that it is installed >> and alive. >> • App sends up to a few hundred KB to the plugin. >> • Plugin sends a few tens of KB back to the app. >> >> This may happen several times a day. >> >> Apple documentation seems to be steering me into Distributed Objects (DO) >> for this. But DO are so complicated! There are vended objects, proxies, >> connection configurations, etc. Looks like I'd have to study it for a week >> or more. Since the web browser talks in JSON-encoded strings anyhow, I >> don't need to send any Cocoa objects; just bytes are fine. >> >> Trying to use what I know, I've been using NSDistributedNotificationCenter >> to send little messages, and when there is significant data, the sender >> writes the data to a temporary file and sends the temporary file's path as >> userInfo in the notification. Upon receiving the notification, the receiver >> reads the file; end of story. >> >> This seems to work OK, except for the minor annoyance of distributed >> notifications always arriving on the main thread and needing to be >> forwarded. I was thinking that maybe I should use NSPort since I've had >> some experience with it, but NSPort documentation advises that "You should >> implement interapplication communication using distributed objects whenever >> possible and use NSPort objects only when necessary." Why is that? >> >> How should I be doing this? > > If the data you are exchanging is fairly lightweight I can recommend shared > memory segments. Decidedly un-Cocoa-like but very fast and reliable. DO is > heavy and distributed notification are unreliable (per the documentation). In > both cases you may need code to ensure reliability. > CFMessagePort is fine too to simply pass data from one process to an other. And it is far more cocoa-like as it is CF based. And as it is a CFRunLoop source, it's easy to decide which thread send and receive messages. -- Jean-Daniel ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
Hi Ken, Thanks for taking a look at my code. On Dec 6, 2010, at 10:35 PM, Ken Thomases wrote: > If you step after the line "TileColumn *allColumns[boardDimension - 1]", > nothing has happened, yet. That's just a declaration. But if for example I declare int test[16], the debugger shows the array "test" with 16 members. That's not happening in the case of my code with allColumns. It's displaying an array of -1 members. > Or am I misunderstanding, and you mean you have stepped repeatedly after that > line, through the loop? Yes, exactly. > Also, what you mean "array remains with a member count of -1"? Which array? > How are you assessing the member count? I'm talking about allColumns. When I inspect allColumns in the debugger after stepping over the line "TileColumn *allColumns[boardDimension - 1]", the debugger shows -1 for the array member count. That's how I'm assessing it. > The allColumns array is a C array, which doesn't know its element count. That doesn't make any sense to me. As mentioned above, if I declare an int array with a member count, it displays as such in the debugger. > The gameColumns array, which I assume is an instance variable of type > NSArray*, is nil until you assign to it in the last statement before the > return. After that assignment, it will be a pointer to an NSArray with > boardDimension elements (or it will have failed with an exception if > something is very wrong, like some element of allColumns is still nil). I understand that the NSArray gameColumns is nil until I assign to it. Unfortunately I am assigning a C array with -1 elements to it because the declaration TileColumn *allColumns[boardDimension - 1] allocates an array with -1 members as displayed in the debugger. >> In the for loop, the line "allColumns[i] = column;" doesn't produce any >> errors but doesn't change the array. > > Doesn't change which array? allColumns? That's basically impossible. How > are you determining this? By looking at allColumns in the debugger. When I look at other code in the debugger that works properly to create and populate an array, I can see objects being added to the array. So why do you say it's impossible to determine this? > Is there any chance that you're trying to debug a release build (or any build > that has optimizations enabled)? An optimized build will be very confusing > to debug. No. It's a debug build. > Finally, something to consider: there's nothing wrong with the technique > you've used of populating a C array and then building an NSArray from that C > array. However, you might consider creating an NSMutableArray that starts > empty and then, during the for loop, each TileColumn is added to it. If you > create the NSMutableArray with +arrayWithCapacity: or -initWithCapacity:, > you'll even avoid reallocation as you add elements to the array. I experimented with doing that unsuccessfully. I am beginning to think that the problem is with having declared boardDimension at the top of my file. I did another experiment and added: int test[boardDimension]; to the file. After stepping over that in the debugger, it showed an array with -1 members. I changed it to int test[16], stepped over that and the debugger showed an array with 16 members. Am I misunderstanding how the debugger works? If not, why is it now working to declare boardDimension at the top of the file?___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
On 7 Dec 2010, at 10:11 AM, Dennis wrote: > But if for example I declare int test[16], the debugger shows the array > "test" with 16 members. That's not happening in the case of my code with > allColumns. It's displaying an array of -1 members. It's possible that the debugger simply doesn't know how to discover the length of a variable-length C array (it's a relatively recent addition to the language). Have you tried changing the array size to a literal 15? The array should have the declared size anyway (and asking the debugger to shoe element 3, etc., should still work) even if the debugger doesn't know its size. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
On Dec 7, 2010, at 10:11 AM, Dennis wrote:
> On Dec 6, 2010, at 10:35 PM, Ken Thomases wrote:
>
>> If you step after the line "TileColumn *allColumns[boardDimension - 1]",
>> nothing has happened, yet. That's just a declaration.
>
> But if for example I declare int test[16], the debugger shows the array
> "test" with 16 members. That's not happening in the case of my code with
> allColumns. It's displaying an array of -1 members.
There is a big difference in declaring an array with an integer literal and
declaring one with an integer variable. In the former, the compiler knows at
compile time what is the size of the array. This is not so in the case of the
latter, which means that the debugger also doesn't know the size of the array.
A smarter debugger might be able to know this, but gdb is not a smarter
debugger. This is not a Cocoa thing; it's just part of the C language. In fact,
variable-length arrays weren't even supported in C until relatively recently
(C99?).
>> Finally, something to consider: there's nothing wrong with the technique
>> you've used of populating a C array and then building an NSArray from that C
>> array. However, you might consider creating an NSMutableArray that starts
>> empty and then, during the for loop, each TileColumn is added to it. If you
>> create the NSMutableArray with +arrayWithCapacity: or -initWithCapacity:,
>> you'll even avoid reallocation as you add elements to the array.
>
> I experimented with doing that unsuccessfully.
You should keep experimenting; using Foundation arrays is usually preferable to
using C arrays for most circumstances, unless speed is really important, and
you won't know if that's the case until you start profiling.
Anyway, this should get you going:
gameColumns = [[NSMutableArray alloc] initWithCapacity:boardDimension];
for (.) {
[gameColumns addObject:[[[TileColumn alloc] init] autorelease];
}
Note the autorelease of the tilecolumn. You need this because NSMutableArrays
will claim ownership of their objects, so you need to relinquish your own
ownership or else you'll have a memory leak.
--
Dave Carrigan
d...@rudedog.org
Seattle, WA, USA
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: OS X Desktop
On Dec 7, 2010, at 8:39 AM, Charlie Dickman wrote: > Thanks Seth. I'll look into AppleScript. Can't you write to the .DS_Store > file with root privilege? You don't need root, you need to force Finder to re-read the file, which you can't do unless you were to quit it, rewrite the file, and relaunch Finder. Either way, it's more likely to break and is more disruptive to the user than AppleScript. -- Seth Willits ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
performSelectorOnMainThread fails second time through
Hi, I'm trying to debug a really odd issue. I have a method that I run in a separate thread, and right at the end of the method, I call [anObject performSelectorOnMainThread:@selector(myMethod:) withObject:someData waitUntilDone:YES]; This works fine the first time I run it. The second and subsequent times I try it, the message never arrives at its destination. I have checked and the destination object is exactly the same, it still has a retain count of 1, and if I ask respondsToSelector: it says yes. Any ideas what else I should look at to try to track down the cause of this? Thanks Gideon ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Another data point on this: if I use waitUntilDone:NO, it doesn't even work the first time through. Responds to selector still returns YES. There are no error messages. On 08/12/2010, at 6:05 AM, Gideon King wrote: > Hi, I'm trying to debug a really odd issue. I have a method that I run in a > separate thread, and right at the end of the method, I call > > [anObject performSelectorOnMainThread:@selector(myMethod:) > withObject:someData waitUntilDone:YES]; > > This works fine the first time I run it. The second and subsequent times I > try it, the message never arrives at its destination. I have checked and the > destination object is exactly the same, it still has a retain count of 1, and > if I ask respondsToSelector: it says yes. > > Any ideas what else I should look at to try to track down the cause of this? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
On Dec 7, 2010, at 2:19 PM, Gideon King wrote: > Another data point on this: if I use waitUntilDone:NO, it doesn't even work > the first time through. > > Responds to selector still returns YES. There are no error messages. Is something tying up the main thread? It won’t execute your request if it doesn’t get to the end of its run loop. Charles___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Interprocess Communication (IPC) : Which "technology" to use?
Thank you. I found an example of shared memory segments here:
http://software.itags.org/software-mac-os/235083/
(Looks like the original server of this page is down.)
And an example of CFMessagePort here:
http://www.omnigroup.com/mailman/archive/macosx-dev/2001-June/028795.html
Given what I know, the CFMessagePort looks like a faster solution for me. Very
lean and just what you'd expect. The server names the port and defines a
callback which returns a response. The client names the same port when sending
a request and the response is returned by reference. I updated Douglas
Davidsons' year-2001 code given in there, to remove compiler warnings.
So it looks like I get to go another year without tackling Distributed Objects.
Whew!
To view the demo, compile and run the Server, then compile and run the Client.
/*** CFMessagePortDemoServer.c ***/
#include
CFStringRef kPortName = CFSTR("MyCFMessagePortDemo") ;
CFDataRef myCallBack(
CFMessagePortRef local,
SInt32 msgid,
CFDataRef data,
void *info) {
char *message = "I'm at your service!";
UInt8 *bytes = (UInt8*)message ;
int dataLength = strlen(message) + 1;
CFDataRef returnData = CFDataCreate(
NULL,
bytes,
dataLength);
printf(
"Received data from a client: %s\n",
CFDataGetBytePtr(data));
// From CFMessagePortCallBack documentation, we return the
// "data to send back to the sender of the message. The system
// releases the returned CFData object."
return returnData;
}
int main() {
CFMessagePortRef local = CFMessagePortCreateLocal(
NULL,
kPortName,
myCallBack,
NULL,
false);
CFRunLoopSourceRef source = CFMessagePortCreateRunLoopSource(
NULL,
local,
0);
CFRunLoopAddSource(
CFRunLoopGetCurrent(),
source,
kCFRunLoopDefaultMode);
// The following will block as long as message port is still valid
// and source remains on the run loop
CFRunLoopRun();
CFRelease(local);
}
Also, quoting from Douglas' original comments,
"As for using it within a Cocoa run loop--by which I assume you mean the main
run loop in an AppKit-based application--that is even easier. In that case,
the server doesn't run the run loop itself, it just adds the source to the
current run loop and lets the AppKit run it…"
/*** CFMessagePortDemoClient.c ***/
#include
CFStringRef kPortName = CFSTR("MyCFMessagePortDemo") ;
int main() {
CFMessagePortRef remote = CFMessagePortCreateRemote(
NULL,
kPortName
);
char *message = "Hello, I'm a client!";
UInt8 *bytes = (UInt8*)message ;
int dataLength = strlen(message) + 1;
CFDataRef data, returnData = NULL;
data = CFDataCreate(NULL,
bytes,
dataLength);
SInt32 result = CFMessagePortSendRequest(remote,
0,
data,
1,
1,
kCFRunLoopDefaultMode,
&returnData) ;
if (
(result == kCFMessagePortSuccess)
&&
(returnData != NULL)
) {
printf(
"Received data from the server: %s\n",
CFDataGetBytePtr(returnData));
CFRelease(returnData);
}
CFRelease(data);
CFRelease(remote);
}
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Good thought thanks Charles, but everything else seems to be responding as usual. No spinning wheel, still able to do other things in the UI. Main thread trace says: #0 0x9546b0fa in mach_msg_trap () #1 0x9546b867 in mach_msg () #2 0x9212d37f in __CFRunLoopRun () #3 0x9212c464 in CFRunLoopRunSpecific () #4 0x9212c291 in CFRunLoopRunInMode () #5 0x92442f58 in RunCurrentEventLoopInMode () #6 0x92442d0f in ReceiveNextEventCommon () #7 0x92442b94 in BlockUntilNextEventMatchingListInMode () #8 0x96aa578d in _DPSNextEvent () #9 0x96aa4fce in -[NSApplication nextEventMatchingMask:untilDate:inMode:dequeue:] () #10 0x96a67247 in -[NSApplication run] () #11 0x96a5f2d9 in NSApplicationMain () #12 0x2d98 in main (argc=1, argv=0xb7d0) at /Users/gideon/Development/CADisplay/main.m:13 which I believe just means that it's idling, waiting for the next event. I tried running the whole thing without using a separate thread for that method, and it worked fine both the first and second time. Gideon On 08/12/2010, at 6:33 AM, Charles Srstka wrote: > On Dec 7, 2010, at 2:19 PM, Gideon King wrote: > >> Another data point on this: if I use waitUntilDone:NO, it doesn't even work >> the first time through. >> >> Responds to selector still returns YES. There are no error messages. > > Is something tying up the main thread? It won’t execute your request if it > doesn’t get to the end of its run loop. > > Charles ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: Problem instantiating an array
On Dec 7, 2010, at 12:11 PM, Dennis wrote: > Thanks for taking a look at my code. Sure. > On Dec 6, 2010, at 10:35 PM, Ken Thomases wrote: > >> If you step after the line "TileColumn *allColumns[boardDimension - 1]", >> nothing has happened, yet. That's just a declaration. > > But if for example I declare int test[16], the debugger shows the array > "test" with 16 members. That's not happening in the case of my code with > allColumns. It's displaying an array of -1 members. As others have pointed out, that doesn't say anything about the code, it says something about the debugger. >> The allColumns array is a C array, which doesn't know its element count. > > That doesn't make any sense to me. As mentioned above, if I declare an int > array with a member count, it displays as such in the debugger. The debugger may know the element count of a fixed-size C array, but the array itself is a "dumb" data structure. You can't query a C array for its element count in code. Also, the number of elements in the array is (potentially) a different question than the capacity of the array. You seemed to be saying that you were expecting the array to know or express how many elements you had assigned into it. It doesn't know that. That is, even if the debugger knew the size of allColumns, that wouldn't change as you assigned things into it. It would be a fixed size and only the contents of each position in it might change. So, stepping through your loop expecting the debugger to say "now there's 1 element in the array, now there are 2 elements, etc.", which is what it seemed your were saying, is unreasonable. >> The gameColumns array, which I assume is an instance variable of type >> NSArray*, is nil until you assign to it in the last statement before the >> return. After that assignment, it will be a pointer to an NSArray with >> boardDimension elements (or it will have failed with an exception if >> something is very wrong, like some element of allColumns is still nil). > > I understand that the NSArray gameColumns is nil until I assign to it. > Unfortunately I am assigning a C array with -1 elements to it because the > declaration TileColumn *allColumns[boardDimension - 1] allocates an array > with -1 members as displayed in the debugger. The expression you used to assign to gameColumns read boardDimension elements from the C array. So, it's simply not true that you were "array with -1 elements to it". There's no such thing as a C array with negative elements. >>> In the for loop, the line "allColumns[i] = column;" doesn't produce any >>> errors but doesn't change the array. >> >> Doesn't change which array? allColumns? That's basically impossible. How >> are you determining this? > > By looking at allColumns in the debugger. When I look at other code in the > debugger that works properly to create and populate an array, I can see > objects being added to the array. Well, you see object pointers being assigned into the positions within the array. You can't "add" to a C array. > So why do you say it's impossible to determine this? I was saying that it's impossible for the statement "allColumns[i] = column;" to not affect the array, assuming 'i' is within the array's bounds. Regards, Ken ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
iOS when is my app launched
How do I find the date and time when my application launched? I've done this before on OS X, but it was a while ago and I've forgotten how. :) Thanks so much! Jason ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: label color
On Sun, 5 Dec 2010 15:47:32 -0700, Nick Zitzmann said: >No. The only parts of Carbon that have gone away are the obsolete APIs >(FSSpec, Internet Config, QuickDraw, etc.) as well as HIView. The rest >of Carbon isn't going anywhere. There are still a number of OS features >that can only be accessed by using Carbon. Funny that you mention Internet Config. Oddly, none of it is even marked deprecated. -- Sean McBride, B. Eng s...@rogue-research.com Rogue Researchwww.rogue-research.com Mac Software Developer Montréal, Québec, Canada ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
I have completely commented out all the actions of the method that is executing in the thread. The only things it does are to create an autorelease pool, call a simple performSelectorOnMainThread, and release the pool. There is nothing else going on that I can see, and yet it still fails as per my previous post. If I have multiple performSelectorOnMainThread calls with wait until completed set to YES, they all work the first time the thread is run, and if set to NO, none of them work. I have taken the threading part of the application and created a little test application with the same flow, and it always works correctly in the test application. Therefore I would conclude that there is nothing I am doing specifically in the threading portion (which is pretty simple anyway), but something else in the program is upsetting it. I have absolutely no idea where else to look. As far as I can tell, the application is just sitting there idle in the main loop waiting for input, so it should execute straight away. I'm not sure what to read into the fact that the first time I call the performSelectorOnMainThread it works if I say for it to wait for completion, but doesn't work if I say not to. I tried running the application in Instruments using the thread states tool, and although I couldn't find any documentation on how to interpret the results, what I believe to be the main thread on both my original application and my mini test application seem to have very similar thread state changes, and what I believe to be the main thread is in "On run queue" state in both cases. I really am at a loss as to where to look next, so would appreciate any suggestions. To restate the issue, when I run a method in a thread and call performSelectorOnMainThread, it works the first time only if I say to wait for completion, and never works if I say not to wait for completion. Any ideas? Thanks Gideon ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
> Any ideas? Post your code! ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Umm, as I mentioned, I have stripped back the multithreading part to virtually
nothing and taken that code to a test application and it works there, and the
rest of the application is 23,000 lines of code, which I can't post.
In my main application, the following still fails (workerThread is an instance
variable):
- (void)execute:(AADataForLayoutProcess *)data {
[self cancel];
self.currentDataForLayout = data;
workerThread = [[NSThread alloc] initWithTarget:self
selector:@selector(doLayout:) object:self.currentDataForLayout];
[workerThread start];
}
- (void)cancel {
[workerThread cancel];
[workerThread release];
workerThread = nil;
}
- (void)doLayout:(id)data {
NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
[self innerExecute:(AADataForLayoutProcess *)data];
[pool release];
}
- (void)innerExecute:(AADataForLayoutProcess *)data {
[self performSelectorOnMainThread:@selector(callback) withObject:nil
waitUntilDone:NO];
}
- (void)callback {
NSLog(@"Called back from thread");
}
At this stage, I'm pretty sure that something other than this code is causing
the issue, but have no clue where to look.
Regards
Gideon
On 08/12/2010, at 11:29 AM, Dave Keck wrote:
>> Any ideas?
>
> Post your code!
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Some thoughts: 1. What happens if you specify a different object to receive the message? For example, try [NSApp performSelectorOnMainThread: @selector(terminate:) ... waitUntilDone: YES]. Does your app terminate? 2. Specify waitUntilDone: NO, and after the call to -performSelectorOnMainThread:, call: CFRunLoopStop(CFRunLoopGetMain()); Does that make it work? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Also: a quick peek at the assembly of -performSelectorOnMainThread: reveals that it calls _CFExecutableLinkedOnOrAfter(). Presumably this is to modify its behavior based on what SDK your app links against; perhaps changing your Base SDK has an effect? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Interesting: 1. No. 2. No. Weird... Now I thought it would be interesting to try those calls before the thread was dispatched, and the terminate one worked, but the run loop stop didn't. I also tried the runloop stop one in my mini test application, and it didn't work there either, so am a bit suspicious of that. Should that kill any application? If so, then maybe there is something wrong on my machine which is showing up in different ways with different apps. I have recently installed the latest XCode. Might try this on another computer and see if that makes any difference. I also checked my build settings, and it was set to use the current MacOS as the base SDK (i.e. 10.6), and the architecture was set to be 32 bit universal, so I changed that to native architecture of build machine. That made no difference. So, no luck so far ... thanks for the suggestions though... Gideon On 08/12/2010, at 2:36 PM, Dave Keck wrote: > Some thoughts: > > 1. What happens if you specify a different object to receive the > message? For example, try [NSApp performSelectorOnMainThread: > @selector(terminate:) ... waitUntilDone: YES]. Does your app > terminate? > > 2. Specify waitUntilDone: NO, and after the call to > -performSelectorOnMainThread:, call: > >CFRunLoopStop(CFRunLoopGetMain()); > > Does that make it work? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
I have copied the source code to a different computer, and can confirm that it made no difference - the behaviour is the same. Regards Gideon ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: OS X Desktop
Thanks Seth. I'll look into AppleScript. Can't you write to the .DS_Store file with root privilege? On Dec 7, 2010, at 1:10 AM, Seth Willits wrote: > On Dec 6, 2010, at 3:11 PM, Charlie Dickman wrote: > >> In the days of System 9 there was a thing known as the desktop database and >> a number of applications/utilities that "remembered" the location of things >> on the desktop and would restore a saved configuration on command. > > Heh. I made one. Clean Desk was the first program I ever sold. I made like > $1200 off of that thing. Pretty nice chunk of change when you're 14. :-) > > > >> For some time now I have been trying to find a similar OS X implementation >> that actually works. All of the ones I have tried mess up when trying to >> restore a saved desktop state; some put icons on top of other icons and some >> place the icons off the screen. >> >> I've tried searching the web for a description of where OS X keeps desktop >> icon placement information, how to read it and how to save it with no luck. > > It really hasn't changed. On OS 9 you had to get the position via > AppleScript. It's the same now unless you dig into the private file format of > .DS_Store (which I spent a lng time doing for DMG Canvas), but then you > can only read the file, not write to it, so you still need to use AppleScript > to get/set the file position reliably, unless you're going to force the user > to quit Finder, rewrite the private file, and relaunch Finder. > > > -- > Seth Willits > > > > ___ > > Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) > > Please do not post admin requests or moderator comments to the list. > Contact the moderators at cocoa-dev-admins(at)lists.apple.com > > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/cocoa-dev/3tothe4th%40comcast.net > > This email sent to 3tothe...@comcast.net Charlie Dickman 3tothe...@comcast.net ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
another Core Data and Undo Question
i have a core data app that supports both real time events and user interaction. some of the user interactions can generate an undo stack similar to the following: beginUndoGroup beginUndoGroup core data action core data action endUndoGroup myAction endUndoGroup now some of the real time events can cause the target of myAction to be deleted and i can easily remove this from the undo stack. but that leaves a nested group which is no longer meaningful and it leaves the undo menu action name referring to the deleted myAction is there anyway to delete the entire group (including the nested core data actions)? if so, how? alternatively, can anyone suggest an approach to deal with this situation? thanx, ken p.s. i don't know if its relevant or not, but i have setGroupsByEvent set to NO and i begin an undo group when i'm ready to execute a user action. ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Cocoaheads Lake Forest meets tomorrow, 12/8 at 7pm
CocoaHeads Lake Forest will be meeting on the second Wednesday of the month. We will be meeting at the Orange County Public Library (El Toro) community room, 24672 Raymond Way, Lake Forest, CA 92630 Please join us from 7pm to 9pm on Wednesday, 12/8. Peter Hosey will walk through the code for an application that uses some of the lesser-known APIs available in Mac OS X. As always, details and the upcoming meeting calendar can be found at the cocoaheads web site, www.cocoaheads.org. (Personal note: Google Irvine is hiring again. Ping me or drop by for details.) ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
> Now I thought it would be interesting to try those calls before the thread > was dispatched, and the terminate one worked, but the run loop stop didn't. I > also tried the runloop stop one in my mini test application, and it didn't > work there either, so am a bit suspicious of that. Should that kill any > application? No, it shouldn't make your app terminate - it'll just cause CFRunLoopRunInMode() to return (see the stack trace you posted earlier.) I was thinking this would cause the -performSelectorOnMainThread machinery to notice that there was an enqueued invocation. (Not that this should be necessary, of course.) On my system, -performSelectorOnMainThread works using a CFRunLoopSource attached to the main thread's run loop, which is signaled after -performSelectorOnMainThread has enqueued an object representing the invocation. I'm thinking -performSelectorOnMainThread isn't working because either the main thread isn't running in a common mode, or something's failing with the CFRunLoopSource. Here are some things to try: 1. Place a breakpoint on the line after the call to -performSelectorOnMainThread. After this is hit, at the GDB prompt, enter the following command: po (void*)CFRunLoopGetMain() This should print a lot of information. One of the first lines printed will mention "current mode =". What mode does it say? 2. In the information printed in the last step, do you see a run loop source mentioned with a callout of "__NSThreadPerformPerform"? (This is the run loop source mentioned previously.) 3. Stop your program, and set a breakpoint on the -performSelectorOnMainThread line. Run your program again. When this breakpoint is hit, before continuing your program, set a breakpoint at CFRunLoopSourceSignal(). Continue your program. The CFRunLoopSourceSignal breakpoint should be hit (directly or indirectly) from within the stack frame of -performSelectorOnMainThread; is this true? ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
> 3. Stop your program, and set a breakpoint on the > -performSelectorOnMainThread line. Run your program again. When this > breakpoint is hit, before continuing your program, set a breakpoint at > CFRunLoopSourceSignal(). Continue your program. The > CFRunLoopSourceSignal breakpoint should be hit (directly or > indirectly) from within the stack frame of > -performSelectorOnMainThread; is this true? Oh - CFRunLoopWakeUp() should also be called after CFRunLoopSourceSignal(). Set a breakpoint on CFRunLoopWakeUp() too, and make sure it's called from within the -performSelectorOnMainThread: stack frame. Furthermore, printing the first argument to both of these functions when/if their breakpoints are hit would also useful: i386: po *(id*)($ebp+8) x86_64: po $rdi ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
Re: performSelectorOnMainThread fails second time through
Thanks for those excellent suggestions Dave (way above my head, but it's good
to have the instructions to execute).
1. Current mode is kCFRunLoopDefaultMode
2. Nothing beginning with NSThread (same was true of my mini test app)
3. Yes, it broke at both those points, and there was something odd going on
with the wake up - it says current mode is (none). It tried this on the mini
test application, and it gave a mode of kCFRunLoopDefaultMode. So it looks as
if we're getting somewhere in tracking down that there seems to be something
wrong with the program at that point, but I really don't fully understand what
it means or where to go next.
#0 0x92154965 in CFRunLoopSourceSignal ()
#1 0x996c27c1 in -[NSThread _nq:] ()
#2 0x996c243d in -[NSObject(NSThreadPerformAdditions)
performSelector:onThread:withObject:waitUntilDone:modes:] ()
#3 0x996d4adf in -[NSObject(NSThreadPerformAdditions)
performSelectorOnMainThread:withObject:waitUntilDone:] ()
#4 0x00057a6c in -[AALayoutExecutor innerExecute:] (self=0x1a15ee0,
_cmd=0x7fe3f, data=0x1e03c30) at
/Users/gideon/Development/Restored/CADisplay/AALayoutExecutor.m:133
#5 0x000572ca in -[AALayoutExecutor doLayout:] (self=0x1a15ee0, _cmd=0x7fe50,
data=0x1e03c30) at
/Users/gideon/Development/Restored/CADisplay/AALayoutExecutor.m:50
#6 0x996acbf0 in -[NSThread main] ()
#7 0x996acba0 in __NSThread__main__ ()
#8 0x9549885d in _pthread_start ()
#9 0x954986e2 in thread_start ()
{locked = No, signalled = No, valid =
Yes, order = 0, context = {version = 0, info =
0x180c150, callout = __NSThreadPerformPerform (0x996c2bbf)}}
#0 0x92154a5d in CFRunLoopWakeUp ()
#1 0x996c2826 in -[NSThread _nq:] ()
#2 0x996c243d in -[NSObject(NSThreadPerformAdditions)
performSelector:onThread:withObject:waitUntilDone:modes:] ()
#3 0x996d4adf in -[NSObject(NSThreadPerformAdditions)
performSelectorOnMainThread:withObject:waitUntilDone:] ()
#4 0x00057a6c in -[AALayoutExecutor innerExecute:] (self=0x1a15ee0,
_cmd=0x7fe3f, data=0x1e03c30) at
/Users/gideon/Development/Restored/CADisplay/AALayoutExecutor.m:133
#5 0x000572ca in -[AALayoutExecutor doLayout:] (self=0x1a15ee0, _cmd=0x7fe50,
data=0x1e03c30) at
/Users/gideon/Development/Restored/CADisplay/AALayoutExecutor.m:50
#6 0x996acbf0 in -[NSThread main] ()
#7 0x996acba0 in __NSThread__main__ ()
#8 0x9549885d in _pthread_start ()
#9 0x954986e2 in thread_start ()
{locked = false, wakeup port = 0x5a03,
stopped = false,
current mode = (none),
common modes = {type = mutable set, count =
1,
entries =>
1 : {contents =
"kCFRunLoopDefaultMode"}
}
,
common mode items = {type = mutable set,
count = 1,
entries =>
25 : {locked = No, signalled =
Yes, valid = Yes, order = 0, context = {version = 0,
info = 0x180c150, callout = __NSThreadPerformPerform (0x996c2bbf)}}
}
,
modes = {type = mutable set, count = 1,
entries =>
10 : {name =
kCFRunLoopDefaultMode, locked = false, port set = 0x5b03,
sources = {type = mutable set,
count = 1,
entries =>
5 : {locked = No, signalled =
Yes, valid = Yes, order = 0, context = {version = 0,
info = 0x180c150, callout = __NSThreadPerformPerform (0x996c2bbf)}}
}
,
observers = (null),
timers = (null)
},
}
}
Thanks again
Gideon
On 08/12/2010, at 4:19 PM, Dave Keck wrote:
>
> 1. Place a breakpoint on the line after the call to
> -performSelectorOnMainThread. After this is hit, at the GDB prompt,
> enter the following command:
>
>po (void*)CFRunLoopGetMain()
>
> This should print a lot of information. One of the first lines printed
> will mention "current mode =". What mode does it say?
>
> 2. In the information printed in the last step, do you see a run loop
> source mentioned with a callout of "__NSThreadPerformPerform"? (This
> is the run loop source mentioned previously.)
>
> 3. Stop your program, and set a breakpoint on the
> -performSelectorOnMainThread line. Run your program again. When this
> breakpoint is hit, before continuing your program, set a breakpoint at
> CFRunLoopSourceSignal(). Continue your program. The
> CFRunLoopSourceSignal breakpoint should be hit (directly or
> indirectly) from within the stack frame of
> -performSelectorOnMainThread; is this true?
___
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
Re: another Core Data and Undo Question
On Tue, Dec 7, 2010 at 5:14 PM, kvic...@pobox.com wrote: > i have a core data app that supports both real time events and user > interaction. some of the user interactions can generate an undo stack > similar to the following: > > beginUndoGroup > beginUndoGroup > core data action > core data action > endUndoGroup > myAction > endUndoGroup > > now some of the real time events can cause the target of myAction to be > deleted and i can easily remove this from the undo stack. but that leaves a > nested group which is no longer meaningful and it leaves the undo menu > action name referring to the deleted myAction > > is there anyway to delete the entire group (including the nested core data > actions)? if so, how? alternatively, can anyone suggest an approach to deal > with this situation? Nope. Welcome to the wonderful world of undo! Where despite the promise of the API, the framework **really** doesn't like it when you call -[NSUndoManager removeAllActionsForTarget:]. Your delete should be undoable, right? So undoing the delete will make your undo group meaningful again. I hope. That's how it works for us in our non-Core Data apps, anyway. Hopefully Core Data is smart enough to be able to associate undo groups with resurrected managed objects. --Kyle Sluder ___ Cocoa-dev mailing list (Cocoa-dev@lists.apple.com) Please do not post admin requests or moderator comments to the list. Contact the moderators at cocoa-dev-admins(at)lists.apple.com Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
