Pike-devel

pike-devel@lists.lysator.liu.se

2 participants
3046 discussions

mmx.h
by Johan Sundstr�m, Lysator ＠ Pike developers forum 25 Oct '03

25 Oct '03

I just received word from a (more) Gentoo person (than me) that it is worth looking for mmx.h as <asm/mmx.h>, in the pike configure scripts. Just posting a note here to make it less probable that it instantly gets forgotten about. :-)

3 4

CVS question
by Bill Welliver 24 Oct '03

24 Oct '03

I'm doing some with the SMTP code that David Gourdelier submitted to CVS, and I've got a logistics question. There are now a number of relatively different classes inside of that module, and I thought it would benefit from being split into multiple code files. The problem is that I need to delete the file SMTP.pmod and replace it with a directory of the same name. Is that going to create problems for people that already have the tree checked out (don't remember if I've ever done this myself). I won't do this until someone gives the OK on that, lest I create real problems. Bill

3 7

autoconf macros for pike
by Bill Welliver 24 Oct '03

24 Oct '03

I made some changes to the autoconf macro that is used to find a pike. It seems to work a little bit better, though it has one shortcoming I'd like to fix: if you specify a pike that's actually a symlink to another, it won't follow the symlink to determine what the actual filename is. This creates a problem when you install a new version of pike (which is a relatively common occurrence), and things stop working because the version has changed: /usr/local/bin/pike is actually /usr/local/pike/ver/bin/pike. Any thoughts about this? -- Specifically, I made some portability changes and added the ability to check for minimum and maximum pike versions. This macro makes it easier to get pike modules set up, as it sets a number of autoconf substitutions. Example in a configure.in/ac when this file is stored as aclocal.m4 in your project directory: AP_CHECK_PIKE(7.4.1, 7.4.999) to check for a version of pike that's between 7.4.1 and 7.4.999, and make sure it has a C include directory (where the headers and compile tools live). It sets the following variables/substitutions: PIKE - path to pike PIKE_VERSION - version of pike found (ie 7.4.10) PIKE_INCLUDE_DIRS - string to includes (ie -I /usr/local/pike/7.4.10/include/pike) PIKE_MODULE_DIR - path to pike modules directory (ie /usr/local/pike/7.4.10/lib/modules)0 I'd like to get some feedback on this, along with other suggestions for easing the creation of pike modules and applications. Thanks again! Bill -- H. William Welliver / Fairchild Semiconductor Mountain Top Systems email: bill.welliver(a)fairchildsemi.com / phone: 570.474.6761 x4445 cell: 570.262.7151

3 5

Defines and dumped modules
by Johan Sundstr�m, Lysator ＠ Pike developers forum 23 Oct '03

23 Oct '03

Are #ifdefs and similar define-dependent code intentionally stripped for dumped modules? I just recompiled a fresh 7.5 and noticed that I could not use -DHTTP_QUERY_DEBUG until I removed all .o files. :-7

5 9

Proposal-patch for documentation of Sql.pike
by Peter Lundqvist (disjunkt) ＠ Pike (-) developers forum 23 Oct '03

23 Oct '03

Oups. Lost a negation. BTW, I can't seem to find documentation for catch in the modref - where should I look for it? $ cvs diff -u Sql.pike Index: Sql.pike =================================================================== RCS file: /cvs/Pike/7.5/lib/modules/Sql.pmod/Sql.pike,v retrieving revision 1.68 diff -u -r1.68 Sql.pike --- Sql.pike 30 Jul 2003 02:30:33 -0000 1.68 +++ Sql.pike 21 Oct 2003 22:59:05 -0000 @@ -340,9 +340,16 @@ return ({sprintf(query,@args), b}); } -//! Send an SQL query to the underlying SQL-server. The result is returned -//! as an array of mappings indexed on the name of the columns. -//! Returns 0 if the query didn't return any result (e.g. INSERT or similar). +//! Send an SQL query to the underlying SQL-server. +//! Returns +//! @ol +//! @item +//! The result as an array of mappings indexed on the name +//! of the columns. +//! @item +//! 0 if the query didn't return any result (e.g. INSERT or similar). +//! @item +//! Throws an exception if the query fails. //! //! @param q //! Query to send to the SQL-server. This can either be a string with the @@ -359,8 +366,12 @@ //! the variable is used. //! //! @code -//! query("select foo from bar where gazonk=:baz", -//! ([":baz":"value"])) ) +//! mixed err = catch { +//! query("SELECT foo FROM bar WHEREe gazonk=:baz", +//! ([":baz":"value"])); +//! }; +//! if(!intp(err)) +//! werror("An error occured."); //! @endcode //! //! Binary values (BLOBs) may need to be placed in multisets.

7 26

Proposal-patch for documentation of Sql.pike
by Peter Lundqvist (disjunkt) ＠ Pike (-) developers forum 22 Oct '03

22 Oct '03

$ cvs diff -u Sql.pike Index: Sql.pike =================================================================== RCS file: /cvs/Pike/7.5/lib/modules/Sql.pmod/Sql.pike,v retrieving revision 1.68 diff -u -r1.68 Sql.pike --- Sql.pike 30 Jul 2003 02:30:33 -0000 1.68 +++ Sql.pike 21 Oct 2003 22:48:00 -0000 @@ -340,9 +340,16 @@ return ({sprintf(query,@args), b}); } -//! Send an SQL query to the underlying SQL-server. The result is returned -//! as an array of mappings indexed on the name of the columns. -//! Returns 0 if the query didn't return any result (e.g. INSERT or similar). +//! Send an SQL query to the underlying SQL-server. +//! Returns +//! @ol +//! @item +//! The result as an array of mappings indexed on the name +//! of the columns. +//! @item +//! 0 if the query didn't return any result (e.g. INSERT or similar). +//! @item +//! Throws an exception if the query fails. //! //! @param q //! Query to send to the SQL-server. This can either be a string with the @@ -359,8 +366,12 @@ //! the variable is used. //! //! @code -//! query("select foo from bar where gazonk=:baz", -//! ([":baz":"value"])) ) +//! mixed err = catch { +//! query("SELECT foo FROM bar WHEREe gazonk=:baz", +//! ([":baz":"value"])); +//! }; +//! if(intp(err)) +//! werror("An error occured."); //! @endcode //! //! Binary values (BLOBs) may need to be placed in multisets.

2 1

Pike-Doc WiKi
by fomyers＠netwalk.com 21 Oct '03

21 Oct '03

I've been trying to learn Pike. Since I already program in C++, XBasic, and other OOP languages, I have noticed that the same roadblock pops up over and over again when I am working in Pike. I keep having to review the pmod "implementation" files to find clarification for parameters, usage of methods, and class inheritance relationships. I think that this is because the reference manual seems to be a bit thin in a few places. As a way to encourage more input to the documentation, Could we use the existing Wiki as a Documentation repository? That way, as Pike Programmers "discovered" information about modules, classes, and methods, we could save the information, or examples of use in the WiKi... Thanks for your time, I hope this helps. Sincerely, Frank Myers

3 3

Pike Interpreter Lock
by Martin Stjernholm, Roxen IS ＠ Pike developers forum 21 Oct '03

21 Oct '03

That should read "garbage collection", not "protection".

1 0

Pike Interpreter Lock
by fomyers＠netwalk.com 21 Oct '03

21 Oct '03

Hi, For integrity, The garbage collection process requires that the Pike global interpreter lock be held. This implies that all threads hold this lock in their turn while executing. Is this a single (intra-)Process lock, or a cross (inter-) Process lock. To put the question another way, does the lock span process boundries, or just threads within a Pike Process? Will this locking single-thread Pike execution in a Multi-Process environment? Frank

2 1

My_Backend and Select
by Martin Stjernholm, Roxen IS ＠ Pike developers forum 21 Oct '03

21 Oct '03

No, sorry, I used "poll" in the wrong context. It uses poll(2) or select(2), so it will really wait in the timeout interval. It's only if you specify a time limit of 0.0 that you'll get a busy loop. / Martin Stjernholm, Roxen IS Previous text: >2003-10-21 21:20: >Subject: My_Backend and Select >-------------------------------------------------------------------- >Martin, > >I think I understand the process of adding a pike program object and running it >as a backend loop, but it's still a polling loop right? > >Isn't that a high overhead process? Isn't there a way to wait for activity on a >stream or other I/O Object? > >Frank > > > > / Brevbäraren >

1 0

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