Pike-devel October 2007

pike-devel@lists.lysator.liu.se

3 participants
24 discussions

pike repo imported into git revision control system
by Martin Bähr 13 Oct '07

13 Oct '07

hi, since i recently got the opportunity to start using git for a project, i tried to import the pike repository into git too. the initial result can be seen here: http://www.iaeste.at/~mbaehr/pike.git this tree is an attempt to reconstruct the history of the main branches (0.5, 0.6, 0.7/7.0, 7.2, 7.4, 7.6 and 7.7) it does not (yet) contain the extra branches. when recontructing the branches i noticed a problem with the various branches in cvs missing files. ie, the contents at the tag pre75to767split should be identical for Pike/7.6 and Pike/7.7 but they are not. this happens in all branches, after the split, some files seem to have gotten lost. i am not yet sure what the best way is to fix this. maybe copying the file from the branch that still has it into the branch where it is missing, and then redoing the import based on that, not sure... anyways, take a look, and let me know what you think. greetings, martin.

6 55

Re: objc
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 09 Oct '07

09 Oct '07

>It seems so odd that the libFoundation folks would go down their own >path wrt exception handling. Effectively, they've created a third, >incompatible way of doing things. Grumble. Yeah. Maybe they're aiming for a solution that works with old objc implementations, such as the original frontend by Brad Cox, which do not have @throw (that syntax was invented about 7 years after the main language itself)... >I was wondering if I can install libfoundation next to the Apple >runtime stuff. It's probably too much to expect that would be >possible. Perhaps I need to get a virtual machine set up just for this. It should be possible to install it under a different prefix, if it works under MacOS at all. >> Needs some ISO standardization methinks... > >I wonder if that'd be better or worse... you'd likely end up with 1 >spec that specifies all the various behaviors :O At least that would mean you could find all of them in one place.

1 0

Re: objc
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 08 Oct '07

08 Oct '07

The reason that there is a conflict between libobjc and libFoundation is that libFoundation depends on its own implementation "libobjc.so.lf2" (lf = libFoundation). Why libFoundation can't use the regular libobjc, or why the special lf version don't work as a drop-in replacement, I don't know.

2 4

Re: objc
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 08 Oct '07

08 Oct '07

>Right, I understand that, but it's a bit like saying "I want to use >C, but not the standard C library." The Foundation framework is the >basic equivalent of the C library, without it, you can't do anything. No, it's not. The standard C library is specified by the C language specification. Foundation is not specified by the objC language specification, it's specified by NextStep. It's more like saying "I want to use C, but not glib".

2 1

Re: objc
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 08 Oct '07

08 Oct '07

Linking with -objc solved that, but now I got a warning: /usr/bin/ld: warning: libobjc.so.lf2, needed by /usr/lib/gcc/powerpc-linux-gnu/4.1.3/../../../../lib/libFoundation.so, may conflict with libobjc.so.1 Perhaps this can be fixed by removing the dependency on libFoundation? Also, running the program does not work: You must call +[NSProcessInfo initializeWithArguments:count:environment:] in main() (Again this is caused by libFoundation, I think.)

4 6

objc
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 06 Oct '07

06 Oct '07

How do I use the Objective-C bindings? When trying to build a simple example program, I get: chiyo:/tmp% gcc -o test test.m -lFoundation /tmp/ccowH8qL.o: In function `main': test.m:(.text+0x1c0): undefined reference to `really_free_svalue' /tmp/ccowH8qL.o:(.data+0x334): undefined reference to `__objc_class_name_OCPikeInterpreter' collect2: ld returned 1 exit status chiyo:/tmp% Ok, I obviously need to create some library. But how? The only part of the build system which touches "OCPikeInterpreter" appears to be the rule "framework", but making framework does not work: chiyo:~/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc% make framework Compiling OCPikeInterpreter.m make[1]: Entering directory `/home/marcus/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc' make[1]: Leaving directory `/home/marcus/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc' make[1]: Entering directory `/home/marcus/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc' make[1]: Leaving directory `/home/marcus/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc' Linking Pike.framework gcc: Cocoa: No such file or directory gcc: dynamic_lookup: No such file or directory gcc: 7.7: No such file or directory gcc: @executable_path/../Frameworks/Pike.framework/Pike: No such file or directory gcc: unrecognized option '-current_version' make: *** [framework] Error 1 chiyo:~/Pike/7.7/build/linux-2.6.22-12-powerpc-ppc% I think this needs some work before it can go into 7.8.

3 7

Keywords
by Martin Nilsson (Opera Mini - AFK!) ＠ Pike (-) developers forum 06 Oct '07

06 Oct '07

As you can see from the sudden spree of checkins we've had another meeting in Linköping (though the initial objective had nothing to do with Pike, the discussion turned in that direction, and we would be stupid to not take notes...) While many of the items are simply bugfixes or incomplete features, there were a few items that dealt with changes that impact the language on a fundamental level. We revisited the "static" discussion and some good arguments were made both for the functionality of static (as in "moved out one scope") and against it. The consensus we had was to directly implement "protected" to do what "static" does today. Previously "protected" did nothing. This has been committed. As next move we would like to implement "shared" as a new keyword that does the "declare in outter scope" thing and to make "static" as deprecated (giving a warning when used). While on the subject of keywords we thought it would be a good idea to also mark "nomask" as deprecated ("final" does the same job). We also want to add "deprecated" as a keyword to be able to mark non-recommended parts of the interface as such. The idea is to tie deprecated into the type system so that it can not only be used on entire functions, but also on specific arguments. A new and annoyingly long #pragma statement to silence deprecation warnings was suggested.

2 1

Re: GTK2 examples
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 04 Oct '07

04 Oct '07

A snapshot is now available at <http://mc.pp.se/gkdrill.pike>. You need the file /usr/share/edict/kanjidic for the program to work.

1 0

Re: GTK2 examples
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 04 Oct '07

04 Oct '07

It's missing lots of functionality. Mostly in the area of selecting the set of kanji to be quizzed on. But the basic stuff works.

1 0

GTK2 examples
by H. William Welliver III 04 Oct '07

04 Oct '07

I got a request the other day for some examples of using the GTK2 module that go beyond "hello world". I only have GTK(1) code; does anyone else have anything that might satisfy an intrepid developer? Bill

2 1

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Pike-devel October 2007