Pike-devel

pike-devel@lists.lysator.liu.se

3041 discussions

signal_handler changes
by Martin Nilsson (Opera Mini - AFK!) ＠ Pike (-) developers forum 26 Jan '13

26 Jan '13

I get (build dir deleted and reconfigured) Compiling signal_handler.c /home/nilsson/pike/src/signal_handler.c: In function âcheck_signalsâ: /home/nilsson/pike/src/signal_handler.c:858:14: error: invalid storage class for function âsignameâ /home/nilsson/pike/src/signal_handler.c:858:1: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement] /home/nilsson/pike/src/signal_handler.c:870:12: error: invalid storage class for function âsignumâ /home/nilsson/pike/src/signal_handler.c:908:13: error: invalid storage class for function âf_signalâ /home/nilsson/pike/src/signal_handler.c:1009:13: error: invalid storage class for function âf_signumâ /home/nilsson/pike/src/signal_handler.c:1030:13: error: invalid storage class for function âf_signameâ [etc]

2 1

pike --help=kladdkaka
by Hedda () ＠ Pike (-) developers forum 26 Jan '13

26 Jan '13

$ pike --help=kladdkaka Ingredients: 4 eggs 5 dl sugar 2 pinches salt 6 msk cocoa 200 g melted butter 3 dl wheat-flour 2 tsk vanilla sugar 1. Mix everything together 2. Pour into a greased circular form. 3. Bake in 175 degrees C (350 degrees F) for 1 hour. 4. Serve while it is still hot with whipped cream. ******************** It should say "6 tbs / 90 ml cocoa" and "2 tsp/10 ml vanilla sugar" otherwise it's hard to understand for non-Swedish people (Never mind that "kladdkaka" is 100% Swedish, but that's the name :P ).

7 15

overflow
by Arne Goedeke 06 Jan '13

06 Jan '13

I pushed a branch containing "hopefully" standard compliant overflow checks. Its in arne/of. I had to change a few places outside of bignum.h in order to get it all working with inlines. The switch from macro to inlines actually changed semantics, since the INT_TYPE_* macros were actually used both with int and unsigned int. Naturally, the inline functions fail to detect an overflow there, since INT_TYPE is 64 bit and calculations do not overflow. Maybe someone can try to compile and run verify on something other than amd64 with gcc? arne

6 21

git access
by H. William Welliver III 04 Jan '13

04 Jan '13

Can someone grant me the ability to push branches? I've got a few feature branches I'd like to get feedback on but I can't push them because they have slashes in them (I assume it's preferred that I not clutter the branch namespace unnecessarily.) Bill

2 1

More aggressive gc of weak mappings in Pike 7.9
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 01 Jan '13

01 Jan '13

Hi. I've now implemented the new rehash behaviour that we discussed during the conference ~2 months ago. This means that weak mappings that contain the only reference to a reference-counted object will free the object when the hashtable is resized. Note that this only takes care of the most common case of weak mappings. The more complicated cases will still need the full gc. For reference: INDEX INDEX VALUE VALUE HANDLING REFTYPE CONTENT REFTYPE CONTENT ------------------------------------------ STRONG * STRONG * KEEP STRONG * WEAK UNREF KEEP STRONG * WEAK SINGLE FREE STRONG * WEAK MULTIPLE KEEP WEAK UNREF STRONG * KEEP WEAK SINGLE STRONG * FREE WEAK MULTIPLE STRONG * KEEP WEAK UNREF WEAK UNREF KEEP WEAK UNREF WEAK MULTIPLE KEEP WEAK SINGLE WEAK * FREE WEAK * WEAK SINGLE FREE WEAK MULTIPLE WEAK UNREF KEEP WEAK MULTIPLE WEAK MULTIPLE KEEP Legend: STRONG Strong (default) reference. WEAK Weak (set with set_weak_flag()) reference. * Any value. UNREF Value that is not reference counted (eg ints or floats). SINGLE Reference-counted value with a single reference. MULTIPLE Reference-counted value with more than one reference. KEEP Entry to be kept after the rehash. FREE Entry to be freed during the rehash. I would appreciate testing, and any comments about the behaviour (intended or otherwise). Happy New Year! /grubba

1 0

Warnings for unused variables
by Jonas Walld�n ＠ Pike developers forum 31 Dec '12

31 Dec '12

When compiling Pike with clang I get massive amounts of warnings for unused variables. This detracts quite a bit from finding more serious errors and I've got a patch ready to remove them. However, before pushing that I'd like to hear if people have objections to changing declarations like in these examples: - static void exit_memory(struct object *o) + static void exit_memory(struct object *UNUSED(o)) and - static int got_udp_event (struct fd_callback_box *box, int event) + static int got_udp_event (struct fd_callback_box *box, int DEBUGUSED(event))

6 18

Filesystem.Tar()->get_dir() behaves oddly
by H. William Welliver III 20 Dec '12

20 Dec '12

I've been playing with a pluggable filesystem class that works with instances of Filesystem.Base, and have noticed that parts of Filesystem.Tar behave differently from Filesystem.System. In particular, paths expressed by Filesystem.Tar always seem to be full rather than the relative paths returned by Filesystem.System. That seems improper to me, especially if one expects Filesystem.Base()->get_dir() to behave in a manner similar to get_dir(). For example: > object f = Filesystem.Tar("foo.tar"); > f->get_dir(); (1) Result: ({ /* 2 elements */ "/Session.pmod", "/Tools.pmod" }) > f = f->cd("Session.pmod"); // note that the object is perfectly content to deal with relative paths > f->get_dir(); (3) Result: ({ /* 2 elements */ "/Session.pmod/RAMSessionStorage.pike", "/Session.pmod/SQLiteSessionStorage.pike", }) I've got a fix that seems to resolve this, but wanted to check that it's actually a bug (it seems as though it must be, as it would be hard for filesystem objects to be interchangeable if they behave differently in this way). Bill

1 0

Planned outage of pike.lysator.liu.se
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 18 Dec '12

18 Dec '12

It should be up now.

3 2

program names
by H. William Welliver III 11 Dec '12

11 Dec '12

I've got some code that generates pike program structs dynamically, and I was wondering if anyone could explain the mechanism that gives C-level programs their names when viewed from pike. For instance: object x = foo.MyProgram > object_program(x); (6) Result: dynamic_class.m:936 Where foo is a module which has `[] overloaded to generate c-api classes on demand. I assume that an entry would normally be created somewhere in the master, however because I'm bypassing all of that machinery, the necessary entries aren't being created. I assume that I should go back and create them, correct? If so, which tables really need to be populated? Bill

1 0

Assign where compare meant
by Peter Bortas ＠ Pike developers forum 02 Dec '12

02 Dec '12

Coverity picked up a few strange assignments in files/ that I commitet fixes for in 7.9. That they had been missed in the files module made me go hunting for more, and I found some in the signal handler. Does this make sense: diff --git a/src/signal_handler.c b/src/signal_handler.c index c886a86..b0a04bb 100644 --- a/src/signal_handler.c +++ b/src/signal_handler.c @@ -1076,7 +1076,7 @@ void forkd(int fd) do { j = close(i); } while ((j < 0) && (errno == EINTR)); - if ((j < 0) && (errno = EBADF)) num_fail++; + if ((j < 0) && (errno == EBADF)) num_fail++; } while (1) { @@ -1088,7 +1088,7 @@ void forkd(int fd) iov.iov_len = 1; do { i = recvmsg(fd, &msg, 0); - } while ((i < 0) && (errno = EINTR)); + } while ((i < 0) && (errno == EINTR)); if (!i) _exit(0); /* Connection closed, shutdown forkd. */ for (cmsg = CMSG_FIRSTHDR(&msg); cmsg; cmsg = CMSG_NXTHDR(&msg, cmsg)) { int ctrl_fd = NULL; @@ -1101,7 +1101,7 @@ void forkd(int fd) num_fds = (cmsg->cmsg_len - CMSG_LEN(0)) / sizeof(int); do { i = fork(); - } while ((i < 0) && (errno = EINTR)); + } while ((i < 0) && (errno == EINTR)); if (i < 0) { /* Fork failure. */ } else if (i) { (Just in case those are there to acctually force the error to EINTR for some reason that I can't see.)

3 2

Jump to page:

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