httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Allan Edwards <a...@us.ibm.com>
Subject Re: cvs commit: apr/network_io/win32 sendrecv.c
Date Thu, 23 Sep 2004 19:37:01 GMT
Joe Orton wrote:
> On Wed, Sep 22, 2004 at 06:21:31PM -0000, ake@apache.org wrote:
>>  --- core.c	20 Sep 2004 20:12:20 -0000	1.286
>>  +++ core.c	22 Sep 2004 18:21:29 -0000	1.287
>>  @@ -2298,7 +2298,7 @@

>>  -    cmd->server->pathlen = strlen(arg);
>>  +    cmd->server->pathlen = (int)strlen(arg);

> It would surely be better to make pathlen an apr_size_t than add casts
> like this.  Did you try that?

This question is going to come up again so it's good to get agreement
before I make many more commits. The IA64 Windows httpd build spits out
a large number of warnings, over 500, and a fair number of these come
from int/size_t mismatches. Apparently other 64 bit compilers
aren't as picky as the MS compiler, since I haven't seen anyone else
complaining about this. They seem to reasonably assume that we aren't
going to have strings >4Gig long.

The general approach I have been taking is changing locally declared
int's to apr_size_t where feasible but I have avoided changing int's
to apr_size_t in public structures.

Changing public structures may ripple through other code
and brings up concerns about MMN bumps. In this case we
would incur a major bump I believe. Unless a particular
field is resulting in warnings in multiple parts of the code
I'm generally going to opt for the cast.

Sound reasonable?

> BTW, if you commit stuff to both httpd and apr at the same time the CVS
> commit notifications don't go to all the right places, so it's best to
> do each separately.

I noticed that after I committed, next time I'll split the commit.

Thanks, Allan


Mime
View raw message