#2654

2015-03-18 Thread John Williams

NetSurf changes page says, re #2654:

> 1. Create an empty !RunImage file in RISC OS page save apps ˜ 

I have racked what few brains I have, but can see no reason for this.

Creation of a zero length dummy file of a totally inappropriate filetype
(text) and which is not even referred to in the Inventory file.

I suppose there must _be_ a reason.

John

-- 
| John Williams 
| joh...@ukgateway.net

 Names for Soul Band:- Soul Expression *



Re: #2654

2015-03-18 Thread Vincent Sanders
On Wed, Mar 18, 2015 at 08:28:45AM +, John Williams wrote:
> 
> NetSurf changes page says, re #2654:

Had you actually read the commit message in its entirity [1] you would
have discovered that the text clearly explains why this was done. 

It fixes bug #1970 [2] by causing the datestamp of the !RunImage file
to be used for the application creation date instead of the directory
whose metadata is easily changed by filesystem operations.

At least thats what the bug said and it was either implement it or
close it. So I implemented it.

> 
> > 1. Create an empty !RunImage file in RISC OS page save apps ˜ 
> 
> I have racked what few brains I have, but can see no reason for this.
> 
> Creation of a zero length dummy file of a totally inappropriate filetype
> (text) and which is not even referred to in the Inventory file.
> 
> I suppose there must _be_ a reason.
> 
> John
> 
> -- 
> | John Williams 
> | joh...@ukgateway.net
> 
>  Names for Soul Band:- Soul Expression *
> 
> 

[1] 
http://git.netsurf-browser.org/netsurf.git/commit/?id=e7f1ad318f85d54bee0bd3e18e8a86cb6688622c
[2] http://bugs.netsurf-browser.org/mantis/view.php?id=1970

-- 
Regards Vincent
http://www.kyllikki.org/



Re: #2654

2015-03-18 Thread John Williams
In article <20150318085030.gd29...@kyllikki.org>,
   Vincent Sanders  wrote:

> Had you actually read the commit message in its entirity [1]

I have now found out how to do this for myself using the link on the
changes page.

Thank you.

John

-- 
| John Williams 
| joh...@ukgateway.net

 Names for Soul Band:- Soul Requirements *



Re: #2654

2015-03-18 Thread David Pitt
Vincent Sanders, on 18 Mar, wrote:

> On Wed, Mar 18, 2015 at 08:28:45AM +, John Williams wrote:
> > 
> > NetSurf changes page says, re #2654:
> 
> Had you actually read the commit message in its entirity [1] you would
> have discovered that the text clearly explains why this was done.
> 
> It fixes bug #1970 [2] by causing the datestamp of the !RunImage file to
> be used for the application creation date instead of the directory whose
> metadata is easily changed by filesystem operations.
> 
> At least thats what the bug said and it was either implement it or close
> it. So I implemented it.

And it works. Tested on Mac VRPC which does not preserve the directory
timestamp on a move.

-- 
David Pitt



Re: #2654

2015-03-18 Thread Rob Kendrick
On Wed, Mar 18, 2015 at 09:15:30AM +, David Pitt wrote:
> 
> And it works. Tested on Mac VRPC which does not preserve the directory
> timestamp on a move.

Ah, is this a bug in VRPC?  This would explain a lot.  (I've been
experimenting in RPCEmu trying to work out wtf this bug is about;
directory time stamps work as I'd expect them to there, regardless of
any !RunImage or lack of one.)

B.



Re: #2654

2015-03-18 Thread David Pitt
Rob Kendrick, on 18 Mar, wrote:

> On Wed, Mar 18, 2015 at 09:15:30AM +, David Pitt wrote:
> > 
> > And it works. Tested on Mac VRPC which does not preserve the directory
> > timestamp on a move.
> 
> Ah, is this a bug in VRPC?  This would explain a lot.

A bug in VRPC!! How very dare you! 

> (I've been experimenting in RPCEmu trying to work out wtf this bug is
> about; directory time stamps work as I'd expect them to there, regardless
> of any !RunImage or lack of one.)

The bit about !RunImage defining the Application date stamp was new to me,
and it does work.

The underlying issue is how the host treats directory moves, real RISC OS or
for exampler SDFS on the Pi, maintains the original date stamp on a move.

HostFS on VRPC Mac, RPCEmu Mac and VRPC Windows 7 on a directory move assign
the current time at the move.


-- 
David Pitt



Re: #2654

2015-03-18 Thread Rob Kendrick
On Wed, Mar 18, 2015 at 10:25:41AM +, David Pitt wrote:
> The underlying issue is how the host treats directory moves, real RISC OS or
> for exampler SDFS on the Pi, maintains the original date stamp on a move.
> 
> HostFS on VRPC Mac, RPCEmu Mac and VRPC Windows 7 on a directory move assign
> the current time at the move.

Then this does sound like a bug or omission in HostFS, not NetSurf.

B.



Re: #2654

2015-03-18 Thread David Pitt
Rob Kendrick, on 18 Mar, wrote:

> On Wed, Mar 18, 2015 at 10:25:41AM +, David Pitt wrote:
> > The underlying issue is how the host treats directory moves, real RISC
> > OS or for exampler SDFS on the Pi, maintains the original date stamp on
> > a move.
> > 
> > HostFS on VRPC Mac, RPCEmu Mac and VRPC Windows 7 on a directory move
> > assign the current time at the move.
> 
> Then this does sound like a bug or omission in HostFS, not NetSurf.

It is not a NetSurf bug, it is just a helpful work around for foibles
elsewhere.

-- 
David Pitt



Re: #2654

2015-03-18 Thread Tony Moore
On 18 Mar 2015, David Pitt  wrote:
> Rob Kendrick, on 18 Mar, wrote:

[snip]

> > Then this does sound like a bug or omission in HostFS, not NetSurf.
>
> It is not a NetSurf bug, it is just a helpful work around for foibles
> elsewhere.

[only just seen this thread] I was the OP of the feature request, and
did explain the reasoning there.

In RISC OS, if a directory is copied, its original datestamp is loat,
however if a !Runimage is present then the application directory shows
the datestamp of the !Runimage, thus preserving the datestamp of the
original full-save.

Thanks to Vincent for implementing the change, which works fine.

Tony






Re: #2654

2015-03-18 Thread David Pitt
Tony Moore, on 18 Mar, wrote:

> On 18 Mar 2015, David Pitt  wrote:
> > Rob Kendrick, on 18 Mar, wrote:
> 
> [snip]
> 
> > > Then this does sound like a bug or omission in HostFS, not NetSurf.
> >
> > It is not a NetSurf bug, it is just a helpful work around for foibles
> > elsewhere.
> 
> [only just seen this thread] I was the OP of the feature request, and did
> explain the reasoning there.
> 
> In RISC OS, if a directory is copied, its original datestamp is loat,
> however if a !Runimage is present then the application directory shows the
> datestamp of the !Runimage, thus preserving the datestamp of the original
> full-save.

Is this with VRPC?

-- 
David Pitt



Re: #2654

2015-03-18 Thread Rob Kendrick
On Wed, Mar 18, 2015 at 12:15:07PM +, David Pitt wrote:
> > In RISC OS, if a directory is copied, its original datestamp is loat,
> > however if a !Runimage is present then the application directory shows the
> > datestamp of the !Runimage, thus preserving the datestamp of the original
> > full-save.
> 
> Is this with VRPC?

A quick experiment shows that this is not the case in RPCEmu in a RAM
Disc (ie, FileCore), but is the case with HostFS.  (RISC OS 5.)

B.



Re: #2654

2015-03-18 Thread Tony Moore
On 18 Mar 2015, David Pitt  wrote:
> Tony Moore, on 18 Mar, wrote:

[snip]

> > In RISC OS, if a directory is copied, its original datestamp is
> > loat, however if a !Runimage is present then the application
> > directory shows the datestamp of the !Runimage, thus preserving the
> > datestamp of the original full-save.
>
> Is this with VRPC?

No. Here, it's ok on RO 6.20 / RiscPC, also on RO 6.20 / RPCEmu / Win7.

Tony






netsurf and gtk+ build error

2015-03-18 Thread ny
Hi guys!

 I try to build netsurf in the distribution PCLinuxOS, but get the
following error:

--cut---
PKG.CNFG: Curl  (libcurl)enabled
PKG.CNFG: GTK-2 (gtk+-2.0)failed
gtk/Makefile.target:62: *** Unable to find library for: GTK-2 (gtk+-2.0).
Stop.
--cut---


 I extremely wondering! Because GTK library is installed, this is evidenced
by the following commands:

$ pkg-config --modversion gtk+-2.0
2.24.22

$ ls /usr/lib64/pkgconfig/gtk+-2.0.pc
/usr/lib64/pkgconfig/gtk+-2.0.pc

Sources of netsurf took here:
http://download.netsurf-browser.org/netsurf/releases/source-full/netsurf-all-3.3.tar.gz


What to do? Please help.


Re: #2654 / dummy !runimage to date an app directory

2015-03-18 Thread Jim Nagel
Tony Moore wrote:
>> In RISC OS, if a directory is copied, its original datestamp is loat,
>> however if a !Runimage is present then the application directory shows
>> the datestamp of the !Runimage, thus preserving the datestamp of the
>> original full-save.

David Pitt  wrote on 18 Mar:
> Is this with VRPC?

Not VRPC per se -- depends on the version of RiscOS.

I haven't got Ro 4.39 to hand at the moment to check, but I think that 
in the Ro 4 *desktop* when you move a directory its datestamp gets 
changed, but if in Configure Filer you switch off "interactive file 
operations" before doing the move (maybe also if you use *copy on the 
commandline) the datestamp is preserved.  At least I remember using 
this switch-off-interactive trick in the past to keep datestamps.

In Ro 5, however, a directory's datestamp seems to be preserved if you 
move the directory in the normal desktop way.

Can't speak for Ro 6; haven't got it.   Obviously VRPC uses 4 or 6.

Application directories, however, are special.  In any version of 
RiscOS an app dir takes its date from the !Runimage within it.  If 
there is no !Runimage its datestamp behaviour is like any other dir.

Thus in the case of an app without a !Runimage, it's a useful trick to 
stick a dummy !Runimage inside the app directory for the sole purpose 
of producing a meaningful datestamp on the outer directory, so that 
the user can easily tell one version of the app from another.

One example is the !PS3 driver, where John Tytgat and Martin Wuerthner 
have followed exactly that suggestion.

So it makes sense in the case of the Netsurf "Full save" too, for the 
sake of users of (apparently) some versions of the OS.


-- 
Jim Nagelwww.archivemag.co.uk



netsurf and gtk+ build error

2015-03-18 Thread ny
Hi guys!

I try to build netsurf in the distribution PCLinuxOS, but get the
following error:

--cut---
PKG.CNFG: Curl  (libcurl)enabled
PKG.CNFG: GTK-2 (gtk+-2.0)failed
gtk/Makefile.target:62: *** Unable to find library for: GTK-2 (gtk+-2.0).  Stop.
--cut---

I extremely wondering! Because GTK library is installed, this is
evidenced by the following commands:

$ pkg-config --modversion gtk+-2.0
2.24.22

$ ls /usr/lib64/pkgconfig/gtk+-2.0.pc
/usr/lib64/pkgconfig/gtk+-2.0.pc

Sources of netsurf took here:
http://download.netsurf-browser.org/netsurf/releases/source-full/netsurf-all-3.3.tar.gz


What to do? Please help.



Re: netsurf and gtk+ build error

2015-03-18 Thread Vincent Sanders
On Thu, Mar 19, 2015 at 01:26:34AM +0400, ny wrote:
> Hi guys!
> 
> I try to build netsurf in the distribution PCLinuxOS, but get the
> following error:
> 
> --cut---
> PKG.CNFG: Curl  (libcurl)enabled
> PKG.CNFG: GTK-2 (gtk+-2.0)failed
> gtk/Makefile.target:62: *** Unable to find library for: GTK-2 (gtk+-2.0).  
> Stop.
> --cut---


please can we have the output of "make Q=" which should make the build
noisy and give us some more information to work with.


> 
> I extremely wondering! Because GTK library is installed, this is
> evidenced by the following commands:
> 
> $ pkg-config --modversion gtk+-2.0
> 2.24.22
> 
> $ ls /usr/lib64/pkgconfig/gtk+-2.0.pc
> /usr/lib64/pkgconfig/gtk+-2.0.pc
> 
> Sources of netsurf took here:
> http://download.netsurf-browser.org/netsurf/releases/source-full/netsurf-all-3.3.tar.gz
> 
> 
> What to do? Please help.
> 
> 

-- 
Regards Vincent
http://www.kyllikki.org/