Pike-devel

pike-devel@lists.lysator.liu.se

1 participants
3045 discussions

Module namespace
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

I have split the nettle module in two files, and need to share some symbols between them. Is it ok to have non-static functions like "hash_init" and "hash_exit" in a module, or do I need to use some more obfuscated name? Is there any nice convention for the names of the initialization function of Nettle/hash.cmod?

1 0

INIT order
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

If I have an inheritance tree, with INIT methods in each class, in which order will they be invoked when a class is cloned? In particular, when using multiple inheritance class foo { inherit A; inherit M; } can I count i on A's INIT method being executed before M's? For some context, M is a mixin, that works by initializing fields of A. Inheritance tree may look something like M A N | | | | B | | |\ / | C E \ / D where N and M are two different mixins modifying the behaviour of A.

2 2

Inherit in .cmod
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

I'm having some trouble, I get a segmentation fault in f_function_name, which I'm not sure how to debug. First, I'd like to know if the inheritance works correctly, the below is the generated code. If I hade written it by hand, I would have put the ADD_STORAGE after the low_inherit calls. #ifdef PROG_MD5_STATE_ID START_NEW_PROGRAM_ID(MD5_STATE); #else start_new_program(); #endif /* PROG_MD5_STATE_ID */ #ifdef THIS_MD5_STATE md5_state_storage_offset=ADD_STORAGE(struct md5_state_struct); #endif /* THIS_MD5_STATE */ #ifdef inherit_hash_instance_md5_state_defined low_inherit(hash_instance_program, NULL, -1, 0, 0, NULL); #endif /* inherit_hash_instance_md5_state_defined */ #ifdef inherit_nettle_md5_md5_state_defined low_inherit(nettle_md5_program, NULL, -1, 0, 0, NULL); #endif /* inherit_nettle_md5_md5_state_defined */ #ifdef md5_state_event_handler_defined pike_set_prog_event_callback(md5_state_event_handler); #endif /* md5_state_event_handler_defined */ md5_state_program=end_program(); add_program_constant("md5_state",md5_state_program,0);

2 1

.cmod
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

How do I write the following program as a .cmod? class foo { program f; mixed `()(int ...args) { return f(@args); } }

3 4

autoconf
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

Jag kör med autoconf-2.57, vilket verkar fungera modulo diverse varningar från både pike och autoconf. Men nu får jag såna här fel cd /home/nisse/hack/pike/src/post_modules/Nettle && autoconf --localdir=/home/nisse/hack/pike/src autoconf: invalid option --localdir=/home/nisse/hack/pike/src Vad är det meningen att --localdir ska göra?

2 3

nettle.cmod
by Niels M�ller () ＠ Pike (-) developers forum 13 Mar '03

13 Mar '03

These functions seems to never have been checked in. Should I do that? Should they be named to_hex/from_hex, or string2hex/hex2string? / Niels Möller () Previous text: >2002-05-10 22:42: >Subject: nettle.cmod >-------------------------------------------------------------------- >Getting down to the optimal number of classes, and making the best use >of the hacks in nettle's nettle-meta.h. > >But I may be able to solve the problem in a better way once I get >inherit to work. At the moment, my problem is that aes->crypt dies of >bus error... > >And I've had to hack the String.from_hex and to_hex functions, as the >GMP-versions didn't handle leading zeroes right. These are the current >definitions. Do they look ok? I thought array_sscanf(s, "%@2x") would >work, but appearantly sscanf doesn't support @. > >/*! @decl string from_hex(string hex) > *! > *! Convert a string of hexadecimal digits to binary data. > *! > *! @seealso > *! @[to_hex()] > */ >string from_hex(string s) >{ > if (sizeof(s) % 2) > error("Can't have an odd number of digits."); > > return (string)(array_sscanf(s, "%{%2x%}")[0] * ({})); >} > >/*! @decl string to_hex(string hex) > *! > *! Convert a string of binary data to hexadecimal digits. > *! > *! @seealso > *! @[from_hex()] > */ >string to_hex(string s) >{ > return sprintf("%@02x", (array(int)) s); >} > > / Niels Möller () >

8 20

Module testsuites
by Niels M�ller () ＠ Pike (-) developers forum 12 Mar '03

12 Mar '03

How are module testsuites supposed to work? I've added a Nettle subdirectory, and run make force_Autoconfig, and recompiled. In the build dir, build/<system>/post_modules/Nettle, there's no testsuite file. When I run make verify, it complains Failed to read test file "testsuite", errno=2. make: *** [verify] Error 1 and when I run make testsuite, it says make: `../../testsuite' is up to date. Running touch testsuite && touch $src/testsuite && make testsuite seems to create a testsuite, though.

4 5

nettle.cmod
by Niels M�ller () ＠ Pike (-) developers forum 12 Mar '03

12 Mar '03

Checked in now. (The first and third chunk, that is, the middle was appearantly fixed already). / Niels Möller () Previous text: >2002-05-13 08:40: >Subject: nettle.cmod >-------------------------------------------------------------------- >New patch for precompile.pike, which actually seems to work: > >Index: precompile.pike >=================================================================== >RCS file: /cvs/Pike/7.3/bin/precompile.pike,v >retrieving revision 1.64 >diff -u -r1.64 precompile.pike >--- precompile.pike 2002/03/13 15:03:25 1.64 >+++ precompile.pike 2002/05/13 06:38:03 >@@ -38,6 +38,7 @@ > * FIXME: this doesn't quite work > * errname; The name used when throwing errors. > * name; The name used when doing add_function. >+ * inherit class; For PIKECLASS, a single superclass can be provided. > * > * FUNCTION OVERLOADING > * You can define the same function several times with different >@@ -1218,8 +1219,8 @@ > array proto=func[..p-1]; > array body=func[p]; > array rest=func[p+1..]; >- string name=(string)proto[-1]; >- mapping attributes=parse_attributes(proto[p+2..], >+ string name=(string)proto[0]; >+ mapping attributes=parse_attributes(proto[1..], > proto[-1]->file+":"+ > proto[-1]->line); > >@@ -1244,6 +1245,9 @@ > " start_new_program();\n" > })) > })+ >+ (attributes->inherit ? >+ ({ sprintf("\n ADD_INHERIT(%s_program, 0);\n", >+ attributes->inherit) }) : ({}) )+ > IFDEF("THIS_"+upper_case(name), > ({ sprintf("\n %s_storage_offset=ADD_STORAGE(struct %s_struct);\n",name,name) }) )+ > subclass->addfuncs+ > > / Niels Möller () >

3 4

nettle.cmod
by Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum 12 Mar '03

12 Mar '03

[...] * PIKECLASS fnord * attributes; * { * INHERIT bar * attributes; * [...]

1 0

Recursive module dependency
by Ludger Merkens 11 Mar '03

11 Mar '03

Hi again, What exactly is the meaning of the Error Message: Recursive module dependency in 'ModuleName' It occurs only if I write a class in a module, that holds instances of its own class. (a parent variable e.g.) Addidtionally it occurs only with my own master, so my main question is probably what may have gone wrong (probably during lookup) that this behaviour occurs. What did i do in the master? (next to some other stuff) *) change master_file_stat that filenames starting with "/" will be prefixed with some "fakeroot" *) make compile_file,read_include, master_read_file aware of this. I also notice, that in the programs cache are empty instances of prefix/path/to/Modulename but compile complains about path/to/Modulename having "recursive module dependencies". hints are welcome. --- Ludger

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