httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (David Robinson)
Subject Re: Comments on current patches.
Date Thu, 14 Sep 1995 10:11:00 GMT
>> Second, there are a few things I don't understand about the SCO patch.
>> *) It increases the size of per-directory configuration vectors by
>>    adding the DYNAMIC_MODULE_LIMIT which came in with last week's
>>    dynamic modules fix.  I'm not sure that this is necessary at all;
>>    per-directory configurations are generally created *after* all
>>    dynamically loaded modules are present and initialized.  If it
>>    is needed, it should probably be a separate patch.
>Umm. My patch doesn't do that, all it does is add a (void **) to the
>pcalloc(). The DYNAMIC_MODULE_LIMIT bit is in the original, too. Whether it
>is right is another question.

What original? It's not in 0.8.13.

>> *) It narrows the 'port' element of the server_rec structure from
>>    int to short.  Is this the right thing to do?
>Ports are short. If I don't narrow it, I get warnings. It might be more
>correct to make it an unsigned short. Excerpt from sys/netinet/in.h:

The warning being harmless, of course.

>> *) Finally, wrt the change to the Configure script, if you're going
>>    to delete the -s option to egrep, the egrep should be redirected into
>>    /dev/null; otherwise, when there are syntax errors, we'll get
>>    junk output in addition to the error message.
>I have to delete -s coz SCO doesn't support it. But, as I read it (from BSDI's
>man pages), -s doesn't suppress egrep's output, it suppresses errors caused
>by nonexistent or unreadable files.

This is incorrect. -s (on some systems at least) supresses egrep output
except for error messages. (The opposite of what you suggest.

>Since egrep is reading from a pipe, -s is a "no effect" option.

Not true (see above).

>Besides, the "junk" output would be the syntax errors. Rather useful, and not
>junk at all, IMHO.

Wrong; have another look at the script. If this egrep fails then the script
itself generates messages about the syntax errors. So output from this
egrep really is "junk".

> > It would be nice if this could all get sorted out before the vote.
>Sigh. OK, at some point in the nearish, I'll upload 04c..., but not until
>people seem to agree on what it should contain.

Why does the patch define SIGURG as SIGUSR1? Wouldn't it be better to disable
SIGURG completely? (By making the use of it in http_main.c #ifdef SIGURG'ed)

The patch contains a C++ style comment; this needs to be removed.

View raw message