Pike-devel January 2004

pike-devel@lists.lysator.liu.se

11 participants
77 discussions

SSLFile
by Mirar ＠ Pike developers forum 16 Jan '04

16 Jan '04

Also, funny warnings: | In file included from /home/mirar/pike/src/program.c:1214: | program_areas.h: In function `low_add_to_program': | program_areas.h:11: warning: left shift count >= width of type | program_areas.h:11: warning: comparison between signed and unsigned | program_areas.h:11: warning: left shift count >= width of type | program_areas.h:11: warning: comparison between signed and unsigned | program_areas.h:11: warning: left shift count >= width of type | program_areas.h:11: warning: signed and unsigned type in conditional expression | program_areas.h: In function `low_add_to_relocations': | program_areas.h:12: warning: left shift count >= width of type | program_areas.h:12: warning: comparison between signed and unsigned | program_areas.h:12: warning: left shift count >= width of type | program_areas.h:12: warning: comparison between signed and unsigned | program_areas.h:12: warning: left shift count >= width of type | program_areas.h:12: warning: signed and unsigned type in conditional expression | program_areas.h: In function `low_add_to_linenumbers': | program_areas.h:13: warning: left shift count >= width of type | program_areas.h:13: warning: comparison between signed and unsigned | program_areas.h:13: warning: left shift count >= width of type | program_areas.h:13: warning: comparison between signed and unsigned | program_areas.h:13: warning: left shift count >= width of type | program_areas.h:13: warning: signed and unsigned type in conditional expression Dunny when they showed up... / Mirar Previous text: >2004-01-16 22:17: >Subject: SSLFile >-------------------------------------------------------------------- >>> SSLPort.pike:65:Index 'SSLFile' not present in module 'SSL'. > >What's up? xenofarm is totally yellow ^^ > > / Mirar >

1 0

Constant expression evaluation
by Alexander Demenshin 16 Jan '04

16 Jan '04

Hi, What about optimizations of constant expressions in Pike? Say, something like "x = 1 + 2 * 3" will be evaluated during compile time or runtime? I tested a simple case "x = 1/0" and got runtime error, which makes me doubt that constant expressions will be reduced at compile time... Regards, /Al

2 1

Nettle and nettle_* identifiers
by Johan Sundstr�m (Achtung Liebe!) ＠ Pike (-) developers forum 16 Jan '04

16 Jan '04

Regarding a recent checkin: > * 2004-01-15 17:26:41 (grendel) > Pike/7.5/src/post_modules/Nettle/configure.in, (+2/-2) (18 lines) > > nice. nettle 1.7 had md5_init, 1.8 has nettle_md5_init, you can't > win. Isn't there some #define to set to have nettle not prefix its identifiers with nettle_ (if that is what it would take to win)?

2 1

core.txt
by Mirar ＠ Pike developers forum 13 Jan '04

13 Jan '04

Pike/xenofarm finally created a core file that is almost readable, for the random error I get from the benchmarks: http://pike.ida.liu.se/generated/pikefarm/7.5/1654_342/core.txt Would it be possible to get it to get a Pike backtrace too, to check which arguments the division had that got that got it to crash so rudily?

2 3

sslfile
by Bill Welliver 13 Jan '04

13 Jan '04

I just tried using HTTP.Protocols.Server.SSLPort in a recent CVS 7.5, and ran into a bunch of problems. Seems there's been a good bit of work on SSL.sslfile recently. I spent about an hour trying to track the problem down, and think I need to defer to someone who's more familiar with the changed material. It looks like the set accept_callback isn't being called with the sslfile object anymore (which is what was happening before). This (obviously) causes problems with code that wants to use the connection that has been negotiated. I traced the problem back in sslfile, and it looks like the accept callback is getting called with the argument of callback_id, which doesn't appear to be set anywhere in the current code, hence it's just a zero. It also doesn't appear to be set anywhere within sslfile to anything other than zero or one. Can anyone with more experience take a look at this and offer suggestions? Bill

1 0

core.txt
by Mirar ＠ Pike developers forum 13 Jan '04

13 Jan '04

No, because gdb_backtraces() requires a running process. Anyway, I run the thread in gdb and caught it: | THREAD_ID 0x4000 (swapped in): | src/modules/Gmp/mpz_glue.c:2037: `/(0) | -:420: test() | /home/mirar/pike/longlongint/test-install/pike/7.5.15/lib/modules/Tools.pmod/Shoot.pmod/CompileExec.pike:56:perform() It seems like mpz can crash on division by zero? Or does mpz::`/ destroy the stack entry?

1 0

Dependency range
by Martin Nilsson (saturator) ＠ Pike (-) developers forum 12 Jan '04

12 Jan '04

What is the range of the OPT_EXTERNAL_DEPEND? The same process? The same computer? The real problem in the issue is dumping I suppose. As an example getpid is added as efun in two places in Pike, in one of them it is marked as OPT_EXTERNAL_DEPEND and in one it is not, underlining the confusion.

3 2

explode_path
by Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum 12 Jan '04

12 Jan '04

I've been thinking a bit about explode_path(). Despite the fact that I don't like the function anyway, I think it could be made at least slightly useful if the semantics was changed so that combine_path(@explode_path(X)) == combine_path(X) (I think this is how things originally worked, but since then combine_path has been updated to behave more logically. :) For UNIX, the only change would be that explode_path("/foo/bar") would return ({ "/", "foo", "bar" }) instead of ({ "", "foo", "bar" }). For AmigaOS and possibly W*ndows, the impact would be a bit larger. The problem is of course that code which does f(explode_path(X))*"/" will break. OTOH, it doesn't seem like the docs say that this should work in the first place... Maybe the old behaviour should be retained by #pike 7.4...

4 4

#include
by Credence Shanden 12 Jan '04

12 Jan '04

Well this is my first attempt with PIKE, although I am familiar with the LPC language some things seem to be a bit different in PIKE. My problem is that I am trying to include a file but I get the error that the file could not be read. I am assuming that somehow I am typing the directory wrong. I have tried many different ways including typing out the full directory name. I have kernel.pike which is the main pike program running in the base directory and the include works there with #include "includes/kernel.h". I tried this with my login sequence file which is located at /lib/obj/std/login.pike but it says it can not find the file. And for future reference I am sure I will be sending out alot of questions. Thanks all in advance. --------------------------------- Do you Yahoo!? Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes

9 15

icc & nan
by Mirar ＠ Pike developers forum 12 Jan '04

12 Jan '04

My icc (7.1) bugs in comparison of NaN. But Pike has it's own support of NaN, doesn't it? Is there any simple way of turning on that support, so I can write a configure test to turn off using a native NaN if it doesn't work?

2 4

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 January 2004