Pike-devel

pike-devel@lists.lysator.liu.se

3 participants
3048 discussions

Start a nNew thread

INT_TYPE = INT64
by Mirar ＠ Pike developers forum 06 Feb '03

06 Feb '03

Hm, I filled in too few '*'. Note that Ackermann gets faster with 12-byte svalues (can anyone explain that?).

3 4

The Gmp, Math modules and the files module in pike 7.5
by Marek Habersack 05 Feb '03

05 Feb '03

Hey all, The Gmp/ and _math/ directories in src/modules/ both contain .cmod files which requires that the system where pike is compiled (even from exported sources) will require an existing pike before even tpike is linked. I thought that moving them to post_modules/ would be a temporary solution, but then files/stat.o linking fails since it cannot find f_max/f_min which are defined in the _math/math.c file. The other solution is to precompile those files when exporting the sources, but alas - I have no idea how to do that :) thanks, marek

3 8

Any way to access the calling object?
by Alexander Demenshin 05 Feb '03

05 Feb '03

Hi everyone, Is there any way to find out which object [instance] is calling some function? I expected that this_object() will do the work, but regardless of the caller it always returns module where it is written. Actually, what I need is to get exact reference to the instance of the object which calls my function (if there is an instance). backtrace() won't work - I'll get the function name and location in the file, but no reference to calling object... Or, to summarize: --- snip otest.pike --- class A { void method() { some_func(); } } some_func() { write(sprintf("Called from: %O", caller())); } int main() { A a = A(); some_func(); // Should print "Called from: otest()" a->some_func(); // Should print "Called from: A()" and (caller() == a) will be true } --- snip otest.pike --- Regards, /Al

2 1

synchronized
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 04 Feb '03

04 Feb '03

mixed synchronized(Thread.Mutex|mapping m, function f) { mixed r; if(mappingp(m)) m = m->_mutex; object key = m->lock(); r = f(); destruct(key); return r; } mapping foo; void do_something() { synchronized(foo) { work on foo }; } should work already, provided you place a mutex in the mapping under the index "_mutex" (or whatever index you choose to look under in the method "synchronized"). / Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) Previous text: >2003-02-04 19:14: >Subject: synchronized >-------------------------------------------------------------------- >One thing that's nice in Java is the 'synchronized keyword. It would >be nice to have that functionality in Pike since it's easier than >manually using mutex locks (nicer code): > > >synchronized void do_something() { ... } > >and > >mapping foo; >void do_something() { > synchronized(foo) { > work on foo > } >} > >and > >void do_something() { > ... > synchronized { > locked region > } > ... >} > >would be nice. The first one creates an implicit mutex lock for the >entire method, the second one creates a lock bound to a specific >variable (i.e you can have multiple methods "locking" the same >variable) and the third method works just like the first, except for a >specific region rather than the entire method. > >Another possibility is to have synchronized classes: > >synchronized class { > ... >} > >and variable instanses: > >synchronized mapping foo = ([]); > >But the first three versions are IMHO more interesting. > >How hard would this be to implement in pike? I'm guessing "quite >hard", but... :) > > / David Hedbor >

3 2

synchronized
by Peter Bortas ＠ Pike developers forum 04 Feb '03

04 Feb '03

See also: much of the reason for implicit lambda. / Peter Bortas Previous text: >2003-02-04 19:14: >Subject: synchronized >-------------------------------------------------------------------- >One thing that's nice in Java is the 'synchronized keyword. It would >be nice to have that functionality in Pike since it's easier than >manually using mutex locks (nicer code): > > >synchronized void do_something() { ... } > >and > >mapping foo; >void do_something() { > synchronized(foo) { > work on foo > } >} > >and > >void do_something() { > ... > synchronized { > locked region > } > ... >} > >would be nice. The first one creates an implicit mutex lock for the >entire method, the second one creates a lock bound to a specific >variable (i.e you can have multiple methods "locking" the same >variable) and the third method works just like the first, except for a >specific region rather than the entire method. > >Another possibility is to have synchronized classes: > >synchronized class { > ... >} > >and variable instanses: > >synchronized mapping foo = ([]); > >But the first three versions are IMHO more interesting. > >How hard would this be to implement in pike? I'm guessing "quite >hard", but... :) > > / David Hedbor >

1 0

Pike 7.4 on Debian/Sparc - a build problem
by Marek Habersack 03 Feb '03

03 Feb '03

Hey guys, I've got pike failing to build a deb on a sparc64 machine. It segfaults while trying to run precompile.sh using tpike. Is it really necessary to run precompile.sh on exported sources? If not - how can avoid it? And the most important issues: - the original build failure record: http://buildd.debian.org/fetch.php?&pkg=pike7.4&ver=7.4.13-2&arch=sparc&sta… - a fragment of a failed build when just running 'make' in a debian-patched directory on sparc64 cpu : TI UltraSparc II (BlackBird) fpu : UltraSparc II integrated FPU promlib : Version 3 Revision 27 prom : 3.27.0 type : sun4u ncpus probed : 1 ncpus active : 1 Cpu0Bogo : 591.46 Cpu0ClkTck : 0000000011a4695a MMU Type : Spitfire the fragment: precompile: /scratch/grendel/pike7.4-7.4.13/build/linux-2.4.18-sparc64/tpike -DNOT_INSTALLED -DPRECOMPILED_SEARCH_MORE -m/scratch/grendel/pike7.4-7.4.13/build/linux-2.4.18-sparc64/master.pike /scratch/grendel/pike7.4-7.4.13/bin/mktreeopt.pike /scratch/grendel/pike7.4-7.4.13/src/treeopt.in (method=QQ) /scratch/grendel/pike7.4-7.4.13/build/linux-2.4.18-sparc64/precompile.sh: line 203: 15768 Segmentation fault $RUNPIKE $SCRIPT "$@" 1>&5 - a final fragment of strace from the above segfault: 22764 mprotect(0x2693c0, 52, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2693c0, 48, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x296da8, 1144, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 2432, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 2424, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 3376, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 4912, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 4912, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 mprotect(0x2adc98, 4912, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 brk(0) = 0x2ba000 22764 brk(0x2bc000) = 0x2bc000 22764 mprotect(0x2b70a8, 10104, PROT_READ|PROT_WRITE|PROT_EXEC) = -1 EINVAL (Invalid argument) 22764 --- SIGSEGV (Segmentation fault) --- 22764 +++ killed by SIGSEGV +++ - and a compile-time warning (not sure if related): /scratch/grendel/pike7.4-7.4.13/src/program.c: In function store_constant': /scratch/grendel/pike7.4-7.4.13/src/program.c:4508: warning: variable e'might be clobbered by longjmp' or vfork' Right now I'm trying to compile the deb on another sparc machine (will take a while), but I'd appreciate any help on that issue... TIA, marek p.s. if you need access to one of the sparc machines, contact me privately, I think we will be able to arrange something.

3 12

Two questions
by Peta, jo det �r jag ＠ Pike developers forum 03 Feb '03

03 Feb '03

I've run in to some problems and would be grateful for some help. 1. How do you combine pike and C in a module? 2. If I have a program struct, what is the best way to decide what kind of program it is?

5 9

MIME.Message problem with 7.4
by David Gourdelier 03 Feb '03

03 Feb '03

Hi list, There is a problem with Pike 7.4 and disp_params variable in MIME.Message. The following code from Caudium doesn't work with Pike 7.4. I tried to copy 7.2's MIME.pmod to Pike 7.4 MIME/module.pmod and then it was working. The difference between two MIME version is that when it works I have part->disp_params=([ /* 1 element */ "name":"1" ]), part->getdata="Upload to attachments", part->headers=([ /* 1 element */ "content-disposition":"form-data; name=\"1\"" ]) and when it doesn't work, this mapping is empty. FYI, Here is the code (from protocols/http.pike and protocols/http2.pike): case "multipart/form-data": // perror("Multipart/form-data post detected\n"); object messg = MIME.Message(data, request_headers); write("data=%O, request_headers=%O\n", data, request_headers); foreach(messg->body_parts||({}), object part) { write("part->disp_params=%O, part->getdata=%O, part->headers=%O\n", part->disp_params, part->getdata(), part->headers); if(part->disp_params->filename) { variables[part->disp_params->name]=part->getdata(); string fname=part->disp_params->filename; if( part->headers["content-disposition"] ) { array fntmp=part->headers["content-disposition"]/";"; if( sizeof(fntmp) >= 3 && search(fntmp[2],"=") != -1 ) { fname=((fntmp[2]/"=")[1]); fname=fname[1..(sizeof(fname)-2)]; } } variables[part->disp_params->name+".filename"]=fname; if(!misc->files) misc->files = ({ part->disp_params->name }); else misc->files += ({ part->disp_params->name }); } else { if(variables[part->disp_params->name]) variables[part->disp_params->name] += "\0" + part->getdata(); else variables[part->disp_params->name] = part->getdata(); } } break; Thanks for your help. -- David Gourdelier

2 4

pike startup
by Ludger Merkens 01 Feb '03

01 Feb '03

Hello, I got strange differences between the startup of pike 7.4 release 13 and pike 7.3 release 60 pike 7.3.30 doing a line like pike -I server/include/ -M server/libraries/ -P server server/server.pike will start "server.pike" pike 7.4.13 will return with: Pike: Couldn't find script to execute ("/home/balduin/cs133/pike4/steam/-M") pwd is: /home/balduin/cs133/pike4/steam doing pike --include-path=./server/include/ --module-path=./server/libraries/ --program-path=server ./server/server.pike on the other hand will work properly. yet pike --help will report the short forms as still valid --- Ludger

3 4

XPM and sort
by Mirar ＠ Pike developers forum 31 Jan '03

31 Jan '03

I finally figured out why XPM decoding is broken on my system. It uses sort() to sort the colors, but a binary comparison in a quicksearch algorithm. sort() uses strcmp. strcmp doesn't sort in binary order on my system¹. I thought it had something to do with large INT_TYPE, but now I'm convinced so isn't the case. This is the order I get: ... "*Y c #eecc22", "*Z c #ffcc22", "=` c #00dd22", "=. c #11dd22", "=+ c #22dd22", "=@ c #33dd22", "=# c #44dd22", ... Note the '`' before '.'. The bug makes the XPM module not find most of the colors, and thus rendering an almost completely black image. Could we make the Pike sort()'s strcmp compare binary? Or does anyone feel up to fixing the XPM module? (¹ appearantly, my default locale is good enough - the bug only appears if the main testsuite is run before. )

4 15

Jump to page:

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