Pike-devel July 2004

pike-devel@lists.lysator.liu.se

9 participants
24 discussions

external module building and environment
by Bill Welliver 29 Jul '04

29 Jul '04

I was wondering if anyone knew if there was any rationale behind the behavior of pike -x module with respect to environment variables. The current behavior is for any environment variables that were in effect for the original build process to take precedent over any in the current environment. This means that it's not possible to use CPPFLAGS and the like to help configure find new libraries and headers. Obviously, that's rather inconvenient. Shouldn't it be the other way around? Maybe the behavior should be more sophisticated (ie combine them)? Any thoughts? Bill

2 1

OpenBSD and pike...
by Xavier Beaudouin 26 Jul '04

26 Jul '04

Hello, According to http://pike.ida.liu.se/generated/pikefarm/7.7/97_28/verifylog.txt We have : Doing tests in modules/files/testsuite (94 tests) 0: ++++++++++ ++++++++++ ++++++++++ ++++++++++ ++++++++++ 50: ++++++++++ ++++++++++ ++++++++++ +++--+++++ Available fds: 1024 Forking... ok. Doing simple tests. Testing dup & assign. / Testing accept. -\|/-\| Testing uni-directional shutdown on socket Testing uni-directional shutdown on pipe Testing 12 sockets. /-\| Testing 16 sockets. /-\| Testing 20 sockets. /-\|/- Testing 24 sockets. \|/-\| Testing 28 sockets. /-\|/-\| Testing 32 sockets. /-\|/-\| Testing 36 sockets. /-\|/-\|/- Testing 40 sockets. \|/-\|/-\| Testing 44 sockets. /-\|/-\|/-\| Testing 48 sockets. /-\Accept returned 0, errno: 9 64 open fds: 0 - 63 [WATCHDOG] Pike testsuite timeout, sending SIGABRT. Abort (core dumped) *** Error code 134 Gdb says (http://pike.ida.liu.se/generated/pikefarm/7.7/97_28/core.txt) : Core was generated by `pike'. Program terminated with signal 6, Abort trap. #0 0x12793dd in poll () Active threads Backtrace #0 0x12793dd in poll () #1 0x7ddfda0 in _thread_kern_poll () #2 0x7ddf746 in _thread_kern_sched () #3 0x7ddf971 in _thread_kern_sched_state () #4 0x7dd9d73 in poll () #5 0x1c112353 in handle_timeouts (ignored=0x0) at /usr/home/kiwi/xenoclient/pike-7.7/obsd.home.oav.net/buildtmp/Pike- v7.7-snapshot/src/modules/HTTPLoop/timeout.c:192 #6 0x7dd703d in _thread_start () #7 0x1f in ?? () Cannot access memory at address 0xffffffff. This bug is here for ages... for both pike 7.6 and 7.7 (I don't have pike 7.4 on xenofarm even my box is sending data... so I cannot compare... do xenofarm does some "blacklisting ????").. Can I help to fix that ? /Xavier -- Xavier Beaudouin - Unix System Administrator & Projects Leader. President of Kazar Organization : http://www.kazar.net/ Please visit http://caudium.net/, home of Caudium & Camas projects

5 7

Re: Access control
by Marcus Agehall (nu med semester) ＠ Pike (-) developers forum 23 Jul '04

23 Jul '04

Hmm, did I missread you or did you get me wrong? Anyway, private will not allow anyone to inherit and extend, it will allow someone to reimplement. Ie, if you have class Foo { private void a(); } class Bar { inherit Foo; void a(); } then Foo.a is not visible from Bar at all. Only Bar.a is accessible. / Marcus Agehall (nu med semester) Previous text: >2004-07-23 07:37: >Subject: Re: Access control >-------------------------------------------------------------------- >Marcus Agehall (nu med semester) @ Pike (-) developers forum wrote: > >>Come to think of it, private might be right anyway. Of course one may >>subclass and override, but that is only because the private methods >>and classes does not exists (well, they aren't visible anyway) in the >>subclass. If the intention is to create a class that should only be >>accessible from your code and never used by subclasses, I would go for >>private. >> >> >> >Yes, private doesn't sound so bad. What I want is to prevent other code >outside of the module >to instantiate new objects from the class, but of course, it won't hurt >if another module is able to >extend the class by inheriting it, and then do what ever it's up to do. >It's to different use cases, whereas the first would prohibit a "end >user" from abusing the class >(should only get already created objects to work with, not create his >own ones) and the second >is for a class to be able to extend the functionality and present a new >object for the end user to use. > >It's all a matter of defining and make the design apparent how to (and >not to) use the module. >It'll be put on modules.gotpike.org as soon as I have it working with >monger, so feel free to have >a look and come back with feedback :o) > > >>Anyway, the modifiers in pike are somewhat confusing if you try to >>apply knowledge from some other language to them... >> >> >> >Indeed, I use to have the documentation for access control open to have >a quick glance when I >get confused. > >Thanks, >Andreas > > > > / Brevbäraren >

1 0

Re: Access control
by Marcus Agehall (nu med semester) ＠ Pike (-) developers forum 22 Jul '04

22 Jul '04

Come to think of it, private might be right anyway. Of course one may subclass and override, but that is only because the private methods and classes does not exists (well, they aren't visible anyway) in the subclass. If the intention is to create a class that should only be accessible from your code and never used by subclasses, I would go for private. Anyway, the modifiers in pike are somewhat confusing if you try to apply knowledge from some other language to them... / Marcus Agehall (nu med semester) Previous text: >2004-07-22 07:57: >Subject: Re: Access control >-------------------------------------------------------------------- >Marcus Agehall (nu med semester) @ Pike (-) developers forum wrote: > >>Ah, that is true. >> >> / Marcus Agehall (nu med semester) >> >>Previous text: >> >> >>>2004-07-20 16:22: >>>Subject: Access control >>>-------------------------------------------------------------------- >>>private is almost never right, since you can inherit and modify then. >>> >>>/ Martin Nilsson (DivX Networks) >>> >>> >>> > >Thanks, I'll look what I get with different attributes. > >//Andreas > > > / Brevbäraren >

1 0

Access control
by Kaos ＠ Pike developers forum 21 Jul '04

21 Jul '04

Hi, I wonder if there's a way to hide a class so only another object/class from the same module can instatiate new objects of the program. I tried to declare the class as `static', which hides it, well enough. But given a instance of this class makes it possible to get around the lock by doing: program p = object_program( my_obj ); object new_obj = p( args ... ); Any way to prevent this? Not a big deal, really, just a thought. //Andreas

4 4

Re: Module name
by Kaos ＠ Pike developers forum 20 Jul '04

20 Jul '04

Ah, ok. Yes, your guess is correct. I'll move it to modules/Lua and see if I can manage to get the installation to put it in Public/Lang/Lua. Is it only monger who does that, or does it work like that when building with make as well? //Andreas / Kaos Previous text: >2004-07-20 14:33: >Subject: Re: Module name >-------------------------------------------------------------------- >Can you maybe elaborate a little more? I'm not sure I follow... is your >c/cmod file in a subdirectory? You shouldn't have to do that; the >install will properly build the module directory. > >Bill > >On Jul 20, 2004, at 8:05 AM, Kaos @ Pike developers forum wrote: > >> Hi, >> >> How should I go about to make configure pick up my module when >> not placed directly in the modules dir? (i.e. in a subdir like >> Public/Lang/..) > > > > / Brevbäraren >

2 1

Module name
by Kaos ＠ Pike developers forum 20 Jul '04

20 Jul '04

Hi, How should I go about to make configure pick up my module when not placed directly in the modules dir? (i.e. in a subdir like Public/Lang/..)

1 0

Re: cmod and pike objects
by Kaos ＠ Pike developers forum 19 Jul '04

19 Jul '04

Thanks, I'll have a look at it. Got some nice examples from the SDL module amongst others as well.. Btw, I'm halfway through to a usable Lua[1] binding module. Thought I could post it on modules.gotpike.org. Question: where in the namespace would be a suitable place to put it? I've thought about Public.Lang.Lua or simply Public.Lua. Any guidelines regarding this? And, before I do, I guess I should get it to run with 7.7 and monger as well :o) //Andreas [1] http://www.lua.org / Kaos Previous text: >2004-07-14 14:08: >Subject: Re: cmod and pike objects >-------------------------------------------------------------------- >There are some very basic docs on the pike website, but imho they don't >even scratch the surface. > >If you're interested in calling pike functions from a cmod, you might >want to take a look at the System.PAM module, the conversation function >gets called by the cmod using safe_apply, iirc. > >Bill > >On Jul 13, 2004, at 8:45 AM, Kaos @ Pike developers forum wrote: > >> Hi, >> >> Is the documentation (more than the very basics) for cmod's "RTFS" >> only? >> >> I'm trying to figure out how to create Pike objects from my cmod, and >> invoke their functions. >> Is calling functions on existing objects used with safe_apply()? with >> args and result on the stack I guess. >> >> more questions to come, I'm sure.. ;o) >> >> Oh, btw, I'm using Pike 7.4 from CVS. >> >> Thanks > > > > / Brevbäraren >

3 4

Glue for C++ libs
by Marcus Agehall (nu med semester) ＠ Pike (-) developers forum 15 Jul '04

15 Jul '04

If I were to write a glue for a C++ lib, what special considerations must be taken into account? Are there any such modules in Pike right now?

3 4

cmod and pike objects
by Kaos ＠ Pike developers forum 13 Jul '04

13 Jul '04

Hi, Is the documentation (more than the very basics) for cmod's "RTFS" only? I'm trying to figure out how to create Pike objects from my cmod, and invoke their functions. Is calling functions on existing objects used with safe_apply()? with args and result on the stack I guess. more questions to come, I'm sure.. ;o) Oh, btw, I'm using Pike 7.4 from CVS. Thanks

1 0

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 July 2004