[fpc-pascal] RTL debuginfo on GDB 7.4+ and OSX Mavericks
Hello list. I've just upgraded to OSX Mavericks. GDB is no longer installed with xcode5 so I downloaded, compiled and installed GDB 7.6.2 myself. I can debug my own programs, testcases and Lazarus packages if I use Dwarf debuginfo. However, RTL which I compiled with "-gl" doesn't show filenames, procnames and line numbers anymore in the heaptrc output or stacktraces. If I compile RTL with "-gl -gw", GDB crashes with the following messages. Using FPC_2_6 and Laz_1_2 from yesterday. While executing the command: "TGDBMIDebuggerInstruction: "-break-insert +0", " gdb reported: "&"linespec.c:2445: internal-error: void decode_line_full(char **, int, struct symtab *, int, struct linespec_result *, const char *, const char *): Assertion `state->canonical_names[i].suffix != NULL' failed.\nA problem internal to GDB has been detected,\nfurther debugging may prove unreliable."" and The GDB command: "-break-insert +0" did not return any result. The GDB process is no longer running. Following [1] I tried to compile GDB 7.3 or below. All of them failed compiling or installing on Mavericks. Is this a known problem? Is there a workaround to see linenumbers and the like on RTL traces on Mavericks? [1] http://www.sourceware.org/ml/gdb/2012-03/msg00055.html Thanks! Joao Morais ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] RTL debuginfo on GDB 7.4+ and OSX Mavericks
On 12/03/2014 11:27, Joao Morais wrote: However, RTL which I compiled with "-gl" doesn't show filenames, procnames and line numbers anymore in the heaptrc output or stacktraces. If I compile RTL with "-gl -gw", GDB crashes with the following messages. Using FPC_2_6 and Laz_1_2 from yesterday. While executing the command: "TGDBMIDebuggerInstruction: "-break-insert +0", " gdb reported: "&"linespec.c:2445: internal-error: void decode_line_full(char **, int, struct symtab *, int, struct linespec_result *, const char *, const char *): Assertion `state->canonical_names[i].suffix != NULL' failed.\nA problem internal to GDB has been detected,\nfurther debugging may prove unreliable."" Well that points to a fault in GDB. Additionally there may or may not be an issue with FPC and the line info it wrote. As for the fault in GDB (crashing, rather than eoither using the line info or reporting an error), I can not predict at which commands it will be triggered. In Lazarus Options (Tools menu), you can change the "InternalStartBreak" (try Entry or MainAddr). It will lead to an alternative of the "+0" break. But that does not fix the original issue, and gdb may crash at any other point. ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] RTL debuginfo on GDB 7.4+ and OSX Mavericks
On 12 Mar 2014, at 12:27, Joao Morais wrote: Hello list. I've just upgraded to OSX Mavericks. GDB is no longer installed with xcode5 so I downloaded, compiled and installed GDB 7.6.2 myself. I can debug my own programs, testcases and Lazarus packages if I use Dwarf debuginfo. However, RTL which I compiled with "-gl" doesn't show filenames, procnames and line numbers anymore in the heaptrc output or stacktraces. That's because no one has implemented a DWARF lineinfo reader for Mac OS X. There is only one for Stabs. Patches are welcome. Jonas ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] RTL debuginfo on GDB 7.4+ and OSX Mavericks
On Wed, Mar 12, 2014 at 8:01 AM, Jonas Maebe wrote: > > That's because no one has implemented a DWARF lineinfo reader for Mac OS > X. There is only one for Stabs. Patches are welcome. > > Is there DWARF reader for other OS targets? Is it a matter of parsing the executable (and/or being smartenough to read from an external .dsym file, if available) thanks, Dmitry ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] RTL debuginfo on GDB 7.4+ and OSX Mavericks
On 12/03/14 17:35, Dmitry Boyarintsev wrote: On Wed, Mar 12, 2014 at 8:01 AM, Jonas Maebe mailto:jonas.ma...@elis.ugent.be>> wrote: That's because no one has implemented a DWARF lineinfo reader for Mac OS X. There is only one for Stabs. Patches are welcome. Is there DWARF reader for other OS targets? Yes. Is it a matter of parsing the executable (and/or being smartenough to read from an external .dsym file, if available) You cannot read it from the executable on Mac OS X, because the DWARF info is left in the object files (to speed up and to reduce memory requirements for linking). So it's either parsing the locations of all object files and then parsing those (if they still exist), or locating and parsing the .dSYM bundle. Only supporting .dSYM bundles would be fine too, as far as I'm concerned. Jonas ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
[fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
I'm working on a project that utilizes a TSortedCollection to help keep items sorted during processing. There is a need to re-sort the collection by different criteria on-demand as the program runs. (Different sort orders are an integral part of the processing.) Unless I'm missing something, TSortedCollection lacks the ability to change the Compare method after initialization and/or explicitly re-sort the collection on-demand. I was planning on handling each new (re)sort by: 1. Initializing a new TSortedCollection with the desired Compare method 2. Copying the contents from the old collection to the new one 3. Swap collection pointers; destroy the old collection. This works without much penalty because the collection will never get beyond a reasonable size, but the implementation makes me cringe. Is there a more elegant way to do this (without writing my own sort routines for a plain TCollection)? -- Jim Leonard (trix...@oldskool.org) Check out some trippy MindCandy: http://www.mindcandydvd.com/ A child borne of the home computer wars: http://trixter.oldskool.org/ You're all insane and trying to steal my magic bag! ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
Sorry, some Turbo Pascal-isms in my initial mail -- I see that FreePascal's TCollection class has an explicit .Sort method so this helps me re-sort on demand, but it doesn't help me switch the Compare method. So, my question still stands. On 3/12/2014 1:51 PM, Jim Leonard wrote: I'm working on a project that utilizes a TSortedCollection to help keep items sorted during processing. There is a need to re-sort the collection by different criteria on-demand as the program runs. (Different sort orders are an integral part of the processing.) Unless I'm missing something, TSortedCollection lacks the ability to change the Compare method after initialization and/or explicitly re-sort the collection on-demand. I was planning on handling each new (re)sort by: 1. Initializing a new TSortedCollection with the desired Compare method 2. Copying the contents from the old collection to the new one 3. Swap collection pointers; destroy the old collection. This works without much penalty because the collection will never get beyond a reasonable size, but the implementation makes me cringe. Is there a more elegant way to do this (without writing my own sort routines for a plain TCollection)? -- Jim Leonard (trix...@oldskool.org) Check out some trippy MindCandy: http://www.mindcandydvd.com/ A child borne of the home computer wars: http://trixter.oldskool.org/ You're all insane and trying to steal my magic bag! ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
using the example at http://acm.msu.ru/mkoshp/fpc/rtl/objects/tsortedcollection.html may something this way ... Type TMethod = ( method_type_1, method_type_2, method_type_3); PMySortedCollection = ^TMySortedCollection; TMySortedCollection = Object(TSortedCollection) method : TMethod; Function Compare (Key1,Key2 : Pointer): Sw_integer; virtual; procedure setMethod( const m : TMethod); end; Implementation Uses MyObject; Function TMySortedCollection.Compare (Key1,Key2 : Pointer) :sw_integer; begin case method of method_type_1: Compare:=PMyobject(Key1)^.GetField - PMyObject(Key2)^.GetField; method_type_2: Compare:=PMyobject(Key2)^.GetField - PMyObject(Key1)^.GetField; method_type_3: Compare:=PMyobject(Key1)^.GetField - 2 * PMyObject(Key2)^.GetField; end; end; procedure TMySortedCollection.setMethod( const m : TMethod); begin method := m; end; Em 12.03.2014 15:51, Jim Leonard escreveu: > I'm working on a project that utilizes a TSortedCollection to help keep > items sorted during processing. There is a need to re-sort the > collection by different criteria on-demand as the program runs. > (Different sort orders are an integral part of the processing.) > > Unless I'm missing something, TSortedCollection lacks the ability to > change the Compare method after initialization and/or explicitly re-sort > the collection on-demand. I was planning on handling each new (re)sort by: > > 1. Initializing a new TSortedCollection with the desired Compare method > 2. Copying the contents from the old collection to the new one > 3. Swap collection pointers; destroy the old collection. > > This works without much penalty because the collection will never get > beyond a reasonable size, but the implementation makes me cringe. Is > there a more elegant way to do this (without writing my own sort > routines for a plain TCollection)? ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
On 3/12/2014 2:30 PM, Philippe wrote: Type TMethod = ( method_type_1, method_type_2, method_type_3); Ah, clever -- create a Compare that switches based on what I tell it to do. I like that, but as far as I can tell there's no way to tell a TSortedCollection to "resort" -- it's kept sorted only through inserts and deletes. -- Jim Leonard (trix...@oldskool.org) Check out some trippy MindCandy: http://www.mindcandydvd.com/ A child borne of the home computer wars: http://trixter.oldskool.org/ You're all insane and trying to steal my magic bag! ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
2014-03-12 21:56 GMT+01:00 Jim Leonard : > On 3/12/2014 2:30 PM, Philippe wrote: > >> Type >> TMethod = ( method_type_1, method_type_2, method_type_3); >> > > Ah, clever -- create a Compare that switches based on what I tell it to > do. I like that, but as far as I can tell there's no way to tell a > TSortedCollection to "resort" -- it's kept sorted only through inserts and > deletes. > You (or the user) will never need to resotre the first order? I am asking because you could as well simply create one SortedCollection for each sort order and systematically insert each item in all those collections. The overhead would be in the initial loading, but then switching would be instantaneous. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] TProcess on windows: incompatibility between poUsePipes and InheritHandles:=false
El 11/03/14 19:43, Luca Olivetti ha escrit: >> I have no explicit solution for you, but a few links that I hope can >> help you further: >> >> http://blogs.msdn.com/b/oldnewthing/archive/2011/12/16/10248328.aspx > > What surprises me is that it says that handles by default are > non-inheritable It turns out that sockets by default *are* inheritable, but that doesn't really work reliably, and at the same time you *cannot* stop them from being inheritable http://stackoverflow.com/questions/12058911/can-tcp-socket-handles-be-set-not-inheritable Now I know that windows has been designed by Vicky Pollard. Bye -- Luca ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
Re: [fpc-pascal] Elegant way to switch sort mechanism in TSortedCollection?
solution may depend on the way the collection is used. load once? many inserts and deletes? ... the time you have to switch from one sort to another ... and the size of the collection ... you could - store the actual collection to a TMemoryStream, - destroy the collection, - Load then a new collect to be sort with different method ... 2014-03-12 21:56 GMT+01:00 Jim Leonard :Em 12.03.2014 19:06, Frederic Da Vitoria escreveu: >> On 3/12/2014 2:30 PM, Philippe wrote: >> >>> Type >>> TMethod = ( method_type_1, method_type_2, method_type_3); >> >> Ah, clever -- create a Compare that switches based on what I tell it to do. I like that, but as far as I can tell there's no way to tell a TSortedCollection to "resort" -- it's kept sorted only through inserts and deletes. > You (or the user) will never need to resotre the first order? I am asking because you could as well simply create one SortedCollection for each sort order and systematically insert each item in all those collections. The overhead would be in the initial loading, but then switching would be instantaneous. > > -- > Frederic Da Vitoria > (davitof) > > Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org [2] > > ___ > fpc-pascal maillist - fpc-pascal@lists.freepascal.org > http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [1] Links: -- [1] http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal [2] http://www.april.org ___ fpc-pascal maillist - fpc-pascal@lists.freepascal.org http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
