[#1378] differences between Module and Class ? — Mathieu Bouchard <matju@...>

25 messages 2003/08/11
[#1387] Re: differences between Module and Class ? — matz@... (Yukihiro Matsumoto) 2003/08/12

Hi,

[#1442] Re: differences between Module and Class ? — Mathieu Bouchard <matju@...> 2003/08/21

[#1406] _id2ref bug? — Ryan Pavlik <rpav@...>

While debugging some caching code, I've come across a segfault related

22 messages 2003/08/14
[#1407] Re: _id2ref bug? — matz@... (Yukihiro Matsumoto) 2003/08/14

Hi,

[#1413] Re: _id2ref bug? (REPRODUCED, short) — Ryan Pavlik <rpav@...> 2003/08/14

On Fri, 15 Aug 2003 01:57:18 +0900

Re: -Wall cleanup project

From: matz@... (Yukihiro Matsumoto)
Date: 2003-08-07 17:58:58 UTC
List: ruby-core #1367
Hi,

In message "Re: -Wall cleanup project"
    on 03/08/06, Elliott Hughes <ehughes@bluearc.com> writes:

|>  * ignore warnings from variables marked as /* OK */ for example, 
|> 
|>      VALUE result;             /* OK */
|> 
|>    I checked and confirmed they are safe, regardless of warnings.
|
|All the submitter was doing was writing "OK" as Qnil: he wasn't rewriting
|the code to avoid the warning. Isn't it a good idea to tell the compiler you
|think you know what you're doing so that -Wall can be the default? That way
|we'd see when it has something important to say.

Oops, gcc 2.x warnings are like

  "warning: variable `saved_block' might be clobbered by `longjmp' or `vfork'"

which cannot be stopped by initializing.  I screwed up.

|For example, VALUE -- the usual type of 'klass' parameters -- is "unsigned
|long" yet in today's CVS, eval.c:220 has the type of a klass parameter
|defaulting to int. That's the only thing that looks convincingly like a
|mistake, but it gets lost in amongst all the "noise" warnings.

Thank you for the report.  The patch is always welcome.

							matz.

In This Thread

Prev Next