Pike-devel

pike-devel@lists.lysator.liu.se

1 participants
3045 discussions

Re: Operator assignment and strict_types
by Marc Simpson 07 Jul '20

07 Jul '20

Thanks! On Tue, Jul 7, 2020 at 5:00 AM Henrik Grubbström (Lysator) @ Pike (-) developers forum <10353(a)lyskom.lysator.liu.se> wrote: > > > Hi folks, > > Hi Marc. > > It took some time, but... > > > It looks like operator assignment (op=), increment and decrement > > statements aren't subjected to the same typechecks as their more > > explicit equivalents. > > Correct. > > > For example, neither the post-increment nor += statements below warn > > with strict_types enabled, even though they assign values outside of > > foo's restricted int domain: > > > > int(3..3) foo = 3; > > foo++; // no warning > > foo += 1; // no warning > > foo = foo + 1; // warning from strict_types > > > > Similarly, operator assignment on aggregates fails to elicit a warning: > [...] > > Fixed in Pike devel. > > The operator assignment expressions now warn in the same places as > their corresponding explict syntax does. > > The ++ and -- operators have not been modified (yet). > > > I'm guessing this is because the checks provided by the > > F_ASSIGN/F_ASSIGN_SELF case (las.c) only apply to direct assignment; > > op= and friends are presumably not instrumented for strict_types in > > the same manner. > > Correct. > > > Have there been any discussions around addressing this inconsistency? > > Now there has... > > /grubba

1 0

Operator assignment and strict_types
by Marc Simpson 07 Jul '20

07 Jul '20

Hi folks, It looks like operator assignment (op=), increment and decrement statements aren't subjected to the same typechecks as their more explicit equivalents. For example, neither the post-increment nor += statements below warn with strict_types enabled, even though they assign values outside of foo's restricted int domain: int(3..3) foo = 3; foo++; // no warning foo += 1; // no warning foo = foo + 1; // warning from strict_types Similarly, operator assignment on aggregates fails to elicit a warning: array(int) a = ({}); multiset(int) b = (<>); mapping(int:int) c = ([]); // no warnings a += ({ "hey" }); b += (< "hey" >); c += ([ "hey": "there" ]); // warnings from strict_types a = a + ({ "hey" }); b = b + (< "hey" >); c = c + ([ "hey": "there" ]); I'm guessing this is because the checks provided by the F_ASSIGN/F_ASSIGN_SELF case (las.c) only apply to direct assignment; op= and friends are presumably not instrumented for strict_types in the same manner. Have there been any discussions around addressing this inconsistency? Thanks, Marc

3 3

an automap bug
by Mateusz Krawczuk 06 Jul '20

06 Jul '20

Reproduces both on pike 8 from apt repository and pike compiled on master: $ ./pike -v Pike v8.1 release 13 Copyright © 1994-2018 Linköping University Pike comes with ABSOLUTELY NO WARRANTY; This is free software and you are welcome to redistribute it under certain conditions; read the files COPYING and COPYRIGHT in the Pike distribution for more details. $ $ cat automap1.pike int main() { write( "%O\n", combine_path("foo", ({ "abc", "cde" })[*] ) ); } $ $ cat automap2.pike int main() { string s = "foo"; write("%O\n", combine_path(s, ({ "abc", "cde" })[*] ) ); } $ $ ./pike automap1.pike ({ /* 2 elements */ "foo/abc", "foo/cde" }) $ $ ./pike automap2.pike automap2.pike:4:[*] not supported here. Pike: Failed to compile script. $ $ pike -v Pike v8.0 release 498 Copyright © 1994-2017 Linköping University Pike comes with ABSOLUTELY NO WARRANTY; This is free software and you are welcome to redistribute it under certain conditions; read the files COPYING and COPYRIGHT in the Pike distribution for more details. $ pike automap1.pike ({ /* 2 elements */ "foo/abc", "foo/cde" }) $ $ pike automap2.pike automap2.pike:4:[*] not supported here.

2 1

Recursive typedefs and add_constant
by Chris Angelico 29 Jun '20

29 Jun '20

I've run into a weirdness involving a recursive typedef and adding a constant. Example: typedef string|int|array(foo) foo; If compiled on its own, this creates a recursive typedef that can handle arbitrarily-nested arrays of strings and ints. Contrast: typedef multiset(string|int) strint; add_constant("strint", strint); typedef string|int|array(strint) foo; This will reference the added constant, and will allow strings, integers, and multisets of strings/ints, but not recursively. What if both names are the same? constant code = #" typedef string|int|array(foo) foo; foo x = 1.0; "; typedef multiset(string|int) strint; void compile_error(string filename, int line, string msg) {write("%s\n", msg);} int main() { write("Without constant:\n"); catch {compile_string(code, "-", this);}; add_constant("foo", strint); write("\nWith constant:\n"); catch {compile_string(code, "-", this);}; } Is the array of foo supposed to be a recursive reference, or a reference to the constant? And is there a way to force one interpretation or the other? A related problem (or a consequence of this) is that repeatedly recompiling a provider of constants can cause strangenesses, up to and including quadratically-increasing compilation time. Consider this cut-down version of part of one of my apps: constant code = #" #ifdef FAST typedef string|mapping(string:string|_echoable_message)|array(_echoable_message) _echoable_message; typedef _echoable_message echoable_message; #else typedef string|mapping(string:string|echoable_message)|array(echoable_message) echoable_message; #endif typedef echoable_message|function(object,object,string:echoable_message) command_handler; bool assess_command(command_handler f) { object|mapping flags = functionp(f) ? function_object(f) : mappingp(f) ? f : ([]); return flags->ok; } protected void create(string n) { #ifndef FAST2 add_constant(\"echoable_message\", echoable_message); #endif add_constant(\"assess_command\", assess_command); } "; int main() { add_constant("commands", ([])); System.Timer tm = System.Timer(); while (1) { compile_string(code)(); write("Compiled in %.2fs\n", tm->get()); } } Running this with -DFAST uses a different name temporarily, dodging the problem. Using -DFAST2 doesn't add the typedef as a constant, also dodging the problem. But otherwise, it quickly starts taking notable time to run each compilation - each iteration about 3-4 times as long as the previous. The question is: Should echoable_message be using the added constant, or should it be defined recursively? I actually don't know what is correct behaviour here. ChrisA

1 0

`-> and the context argument
by Stephen R. van den Berg 20 Jun '20

20 Jun '20

The docs say this: mixed `->(string index, object|void context, int|void access) Parameter context Context in the current object to start the search from. If UNDEFINED or left out, this_program::this is to be be used (ie start at the current context and ignore any overloaded symbols). Can someone illustrate this with an example which shows how context can differ and how it could be used? -- Stephen.

2 1

pike with -fstack-protector
by william＠welliver.org 12 Jun '20

12 Jun '20

I'm working on getting pike included in pkgsrc, and there's been a question about whether it's possible to get a usable pike with stack smashing protection enabled. When I originally prepared the port, I had to disable SSP on the build. In the intervening years I've forgotten the exact circumstances around why that's necessary. A suggestion was made that this problem has been overcome by newer versions of gcc for certain cases where it had been a problem, but I wanted to get more feedback to see if anyone is/has built and used pike with one of the various -fstack-protector options. Thanks in advance! Bill

1 0

expired pike.lysator.liu.se SSL certificate
by Stanislaw Klekot 12 Jun '20

12 Jun '20

A colleague of mine noticed that the SSL certificate of Pike's home page has expired a week ago. https://pike.lysator.liu.se/ We'd appreciate it getting renewed. -- Stanislaw Klekot

1 0

Any interest in an open source HTTP daemon in Pike? (Spike)
by Stephen R. van den Berg 12 Jun '20

12 Jun '20

I've written (for internal use) an HTTP daemon in Pike, using roughly the following featureset: - Mainly meant to run (custom) pike modules processing requests. - Also serves static content (through a simple config file). - Supports multiple virtual servers, with multiple mountpoints each. - Supports database connection pools. - Supports integrated memory caching. - Uses the Shuffler for all network I/O. - Supports websockets. - Gzips output automatically. - Uses statically precompiled .gz files if available and new enough. - It has been fondly baptised as a "Spike" daemon. - It integrates well with an nginx reverse proxy (and thus I did not add SSL support so far), which then adds HTTP/2 and SSL support. - It only runs on Pike 8.1, not Pike 8.0 (mostly because of the broken Shuffler in 8.0). - The whole source is about 15 KB (5 KB when gzipped). If there is interest, I'll clean up the code a bit more, and make it available via a git repo somehwere. -- Stephen.

2 4

Addressing the outer class
by Stephen R. van den Berg 11 Jun '20

11 Jun '20

class a { int c = 4; class b { int c = 3; void set(int d) { c = d; // FIXME } } void runme() { object y = b(); y->set(12); } } object x = a(); x->runme(); write("a->c %d\n", x->c); If I want this to print 12 (instead of 4), how do I need to tell Pike to access the c-member from b->set() in class a instead of the one in class b? -- Stephen.

1 1

Compiling programs in Pike master with proper error messages
by Stephen R. van den Berg 07 Jun '20

07 Jun '20

I'm using the following function in my spike-webserver: object compileit(string file, void|int quiet) { object prog; if (!quiet) log("Compiling %s", file); if (mixed err = catch(prog = compile_file(file)())) log(LOG_ERR, "Compile error %s", describe_backtrace(err)); return prog; } in order to compile modules. I'd like two things though, which currently don't quite work as I had hoped: a. When there is a compilation problem, I get a backtrace here, but it does not reveal what compilation error occurred (it does not show line number and diagnostics). Any suggestions how to get those? b. I'd like a (list of) custom library directories which I want to include in the namespace, without having to specify them on the commandline using -L options. How can this be accomplished? -- Stephen.

3 2

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