Pike-devel

pike-devel@lists.lysator.liu.se

2 participants
3046 discussions

Protocols.HTTP.Query and 100 Continue
by H. William Welliver 13 Jan '26

13 Jan '26

Hello all, I found myself interacting with a HTTP server that returns 100 Continue responses to HTTP POST requests. The pike HTTP client treats those as the final response, but they’re are meant to be considered intermediate responses. I couldn’t figure out if there was a way to get it to continue past that, and it doesn’t look like there is. The standard says a server shouldn’t send them if they aren’t requested, but also says that clients must treat them as informational and carry on. I have a proposed fix, but since that’s a messing with a pretty widely used bit of code, I wanted to get feedback before I start making changes. I don’t know if there are other similar scenarios that this fix doesn’t cover; I’d be happy to hear feedback on that as well. In Protocols.pmod/HTTP.pmod/Query.pike, following this line: int bytes_read = sscanf(headerbuf,"%s%*[ ]%d%*[ ]%s%*[\r]",protocol,status,status_desc); add: // we don't expect continue responses, but we must deal with them, // and treat them as informational if(status == 100) { int q = sizeof(headerbuf); // find the end of the response message, and move past the newlines present. // since the whole request has been sent already, we just have to wait for the response, // if it isn’t already in the buffer. Either way, there should be trailing line endings, // as those were required to be present for us to have gotten this far. while(q < sizeof(buf)) { if(buf[q] == '\n' || buf[q] == 'r') continue; break; } // reset the buffer to the point just past the continue response and start over. buf = buf[q+1..]; // a better approach might be a do-while loop, but this is a bit clearer, // and we shouldn't find ourselves recursing more than once. return ponder_answer(); } I think this affects all versions of pike (certainly 8.x and 9.x). Thanks in advance! Bill

1 0

Generated (precompiled) function names changing
by Magnus Holmgren, Millnet/Lysator/Debian/Mensa ＠ Pike developers forum 07 Jan '26

07 Jan '26

When building both the current stable 8.0 and 9.0.11, *sometimes* the C code gets regenerated from .cmod (by precompile.pike, I guess) and then the macros in src/modules/Gmp/mpq.c change names from f_mpq_* to f_Gmp_mpq_* (and similarly in other modules), which in that case causes compilation to fail because mpq.cmod contains references to some of the generated function names, e.g. f_mpq_get_int(0) in cast(string). Any idea how this is happening and if it's supposed to happen and the function calls should be changed to match?

2 4

Re: Availability of Pike SRPMs for EL8 and newer builds
by H. William Welliver 11 Dec '25

11 Dec '25

So, as usual my time estimates were a bit off. I’ve cleaned my scripts up a little and have posted them here: https://hg.sr.ht/~hww3/pike_rpm_build I use docker and rpmbuild to generate the RPMS, and I have successfully built and tested packages for EL8 and EL9 on amd64 and aarch64. Things don’t build on EL10 for a couple of reasons; I think some work will be required to get things working properly there (a combination of dependencies being renamed/deleted as well as some build errors probably triggered by the latest versions of gcc. I’ve included the non-working EL10 setup in case anyone wants to take a stab at fixing the problems. The use of docker allows us (hopefully) to have a reproducible, pristine build environment, and also allows RPMS to be built from non-linux platforms without needing to have a linux system set up. There are notes in the repo above that describe how to build the docker build images, and then use them to prepare (S)RPMS, as well as the basic steps needed to support new versions of Enterprise Linux. I did notice a minor problem on Oracle Linux 9 with there being odd dependencies that couldn’t be resolved (Java related, mainly and as per the usual).You can ignore them or better yet, break them into a sub-package. I’ll try to make time to upload my builds later today or tomorrow. Comments/questions/suggestions are welcome, as always. Bill > On Dec 6, 2025, at 9:00 PM, Maksim AbuAjamieh <eng.maksim(a)gmail.com> wrote: > > > Hi Bill, > > Thanks so much for the update and for taking the time to clean things up and make them accessible. > > I’ll look forward to the SRPMs whenever you’re ready to share them, but it is perfectly fine if you would prefer to just share what you already have. I really appreciate the effort you are putting into this. > > Best regards, > Maksim > > On Wed, Nov 19, 2025 at 7:41 AM <william(a)welliver.org <mailto:william@welliver.org>> wrote: >> Hi- >> >> That would be my fault... I've always been so focused on the rpms but >> the srpm is an artifact I seem to forget to upload the SRPM. >> >> I have a spec file that's never been checked into version control, as >> well as some scripts that I've used to produce the RPMs available for >> download... Part of the reason it's always languished on my build system >> is that the spec file needs changing for various el releases, and >> there's a separate dependency installation step required to get the >> builds to work. I didn't want a proliferation of files in the pike >> packaging directory. I will try to clean those up a bit and put them >> somewhere publicly accessible. >> >> Note: pike doesn't seem to build properly on el10, I suspect due to the >> version of gcc in use. The fix is a one-liner, but a patch is necessary >> until a new release is made. >> >> Stand by for an update on this, probably in a day or two. >> >> Best, >> >> Bill

1 0

Pike 9.0.11 (beta 3) released.
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 11 Dec '25

11 Dec '25

The above is now available for download from the Pike site. Please test and enjoy. /grubba

1 0

Re: Playing music using pike and SDL.
by Martin Nilsson (lvl 60) ＠ Pike (-) developers forum 03 Dec '25

03 Dec '25

Were you not able to get any C code to play through SDL either? Does your SDL library have a mixer compiled in it (I think it is optional)?

3 2

compile errors
by Lance Dillon 02 Dec '25

02 Dec '25

At least Pike-v8.0.1916 on Fedora 43 fails in src/modules/HTTPLoop/requestobject.c: Line 693, change: int true=1; setsockopt( a->to->fd, SOL_TCP, TCP_CORK, &true, 4 ); to: int istrue=1; setsockopt( a->to->fd, SOL_TCP, TCP_CORK, &istrue, 4 ); Same with line 796 with false, change to isfalse. true and false are reserved words with c99. Or add -std=gnu90 flag (with gcc at least) so that true and false aren't reserved anymore. Probably changing the code is better. Also, src/modules/_WhiteFish/blobs.c: Change: static void exit_blobs_struct( ); to: static void exit_blobs_struct(struct object *o); linkfarm.c: Change: static void exit_linkfarm_struct( ); to: static void exit_linkfarm_struct(struct object *o); These might have a gcc option to prevent this error (signatures don't match), but I don't know what it is. Just something quick I found compiling for a new system.

2 2

Pike Meet-up 2024-Q2
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 10 Oct '25

10 Oct '25

Hi. At Pike Conference 2017 it was decided that we ought to hold a more informal meeting in Linköping ~ once every quarter. As the 6 month mark from the latest conference is coming up soon, it seems time to announce the second such meeting for 2024: Pike Meet-up 2024-Q2 Date: 2024-05-11 (Sat) Time: 10:00 - afternoon/evening Place: Roxen IS AB Storgatan 46 Linköping Welcome! /grubba

1 4

mmap'ing a device with System.Memory
by Chris Angelico 18 Jan '25

18 Jan '25

Currently this isn't possible - if you attempt to open System.Memory("/dev/some-device") it rejects it due to being unable to query the length of the file. It would be nice to be able to, for example, stick Pike on a Raspberry Pi, and then mmap /dev/gpiomem to manipulate things. Would it be of value to have a way to bypass that length check? I think that that's the only thing stopping it, although I'm not entirely certain. ChrisA

2 1

macOS Homebrew MySQL / MariaDB support
by Bertrand Lupart 04 Oct '24

04 Oct '24

Hello, Looks like Pike’s Homebrew formula is lacking MySQL / MariaDB support : $ brew info pike ==> pike: stable 8.0.1738 (bottled) […] $ pike --features auto_bignum machine_code dynamic_modules threads out-of-band_data double_precision_float Bz2 Crypto.AES.GCM Crypto.ECC.Curve GL GLUT GSSAPI Gettext Gmp Gz HTTPAccept Image.GIF Image.JPEG Image.PNG Image.TIFF Image.XFace Kerberos MIME Nettle PDF.PDFlib Protocols.DNS_SD Regexp.PCRE Regexp.PCRE.Widestring SQLite Standards.JSON System.FSEvents.EventStream WhiteFish Yp Surprisingly, looks like that was removed a while ago : https://github.com/Homebrew/homebrew-core/commit/3d9568bc86dc58c869c6e68a2e… Any reason why this change was made ? Bertrand

2 1

Understand the behaviour of Stdio.Buffer()->match()
by Chris Angelico 08 May '24

08 May '24

I'm a bit confused by this behaviour, and am unsure whether it needs to be better documented, or perhaps is a bug. int main() { object buf = Stdio.Buffer("#44"); write("Matches: %O %O\n", buf->sscanf("#%2[4]"), (string)buf); buf = Stdio.Buffer("#4"); write("Doesn't: %O %O\n", buf->sscanf("#%2[4]"), (string)buf); buf = Stdio.Buffer("#44"); write("Matches: %O %O\n", buf->match("#%2[4]"), (string)buf); buf = Stdio.Buffer("#4"); write("Doesn't: %O %O\n", buf->match("#%2[4]"), (string)buf); } When using sscanf, the successful match returns ({"44"}) and leaves the buffer empty, and the unsuccessful match returns an empty array and leaves the "4" in the buffer. This makes sense; the hash was matched and consumed, but the 2-character requirement meant that nothing was matched. Using match instead produces a somewhat odd result though. If there's enough text in the buffer, it returns "44" and leaves the buffer empty; this corresponds to sscanf's behaviour. In the other case, though, it returns "#%2[4]" (the pattern), and leaves "4" in the buffer. Is this intentional? (If the first character of the buffer isn't a hash, both return 0 and leave the buffer unchanged. This behaviour is also correct. It's only the partial match that seems curious.) ChrisA

3 3

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