Mailing List Archive

Compiling with -Wall
I made some changes to the first apache-pre release to fix some
warnings generated when compiling on NetBSD-current with -Wall
flag for gcc. Some of these warnings are truely anal, but it
often does find some real bugs, and I may have found a couple
in this case. Do with it what you want. There is also some
patches that are specific to BSD4.4, and some specific to
BSD which I think Brian has already caught. Would someone
please look closely at ~185: util.c to make sure that I read
this test correctly?

Looking at the homepage, it looks like we have reached a decision
on "Apache"? If so, I will begin moving forward on a logo.

Also, I attempted to use the "submit a patch" link, but could not
raise steam.com.

My patch is on hyperreal in the incoming directory. (patch.wall)
Re: Compiling with -Wall [ In reply to ]
Well it seems that power is getting real flakey at my
place due to the storm (power keeps dropping). I'll be
moving my script to hyperreal tonight...!@#$% storm.

Cliff
Re: Compiling with -Wall [ In reply to ]
> > Date: Sat, 11 Mar 1995 21:52:51 -0600
> > From: Randy Terbush <randy@zyzzyva.com>
> > I made some changes to the first apache-pre release to fix some
> > warnings generated when compiling on NetBSD-current with -Wall
> > flag for gcc. Some of these warnings are truely anal, but it
> > often does find some real bugs, and I may have found a couple
> > in this case. Do with it what you want. There is also some

>
> I certainly think that these warnings should be fixed, and the coding
> style in general needs improving. However, right now I think we want
> to avoid changing too many source files. Not only would we move further
> away from the NCSA source, but it also makes it a little harder to
> integrate patches with one-another.
>
> The majority of your changes are to remove redundent variables (of which
> there are sadly too many) -- which an optimising compiler would do anyway,
> and to insert parenthesis around assignments in conditional expressions.
> May I suggest that you trim down the patch to include only the bug fixes?

Point well taken. I will trim this and resubmit it.
Re: Compiling with -Wall [ In reply to ]
> X-Uri: http://www.inetnebr.com/zyzzyva
> Date: Tue, 14 Mar 1995 07:15:21 -0600
> From: Randy Terbush <randy@zyzzyva.com>
>
> > I certainly think that these warnings should be fixed, and the coding
> > style in general needs improving. However, right now I think we want
> > to avoid changing too many source files. Not only would we move further
> > away from the NCSA source, but it also makes it a little harder to
> > integrate patches with one-another.
> >
> > The majority of your changes are to remove redundent variables (of which
> > there are sadly too many) -- which an optimising compiler would do anyway,
> > and to insert parenthesis around assignments in conditional expressions.
> > May I suggest that you trim down the patch to include only the bug fixes?
>
> Point well taken. I will trim this and resubmit it.
>
> Aw, drat. The content-negotiation stuff really does need another
> pass, and I was looking forward to finally being able to use -Wall
> while debugging it. (I suppose I could have the Makefile turn it on
> only for http_mime_db.c for the time being).
>
> Brian and Cliff have both been doing integration --- has this been a
> problem so far?

I don't *have* to remove this. :-) I have been testing my changes locally,
and with exception of the one *nasty* bug I created that Brian pointed
out, I don't think any of these changes caused other problems.
Re: Compiling with -Wall [ In reply to ]
X-Uri: http://www.inetnebr.com/zyzzyva
Date: Tue, 14 Mar 1995 07:15:21 -0600
From: Randy Terbush <randy@zyzzyva.com>

> I certainly think that these warnings should be fixed, and the coding
> style in general needs improving. However, right now I think we want
> to avoid changing too many source files. Not only would we move further
> away from the NCSA source, but it also makes it a little harder to
> integrate patches with one-another.
>
> The majority of your changes are to remove redundent variables (of which
> there are sadly too many) -- which an optimising compiler would do anyway,
> and to insert parenthesis around assignments in conditional expressions.
> May I suggest that you trim down the patch to include only the bug fixes?

Point well taken. I will trim this and resubmit it.

Aw, drat. The content-negotiation stuff really does need another
pass, and I was looking forward to finally being able to use -Wall
while debugging it. (I suppose I could have the Makefile turn it on
only for http_mime_db.c for the time being).

Brian and Cliff have both been doing integration --- has this been a
problem so far?

rst
Re: Compiling with -Wall [ In reply to ]
On Tue, 14 Mar 1995, Robert S. Thau wrote:
> X-Uri: http://www.inetnebr.com/zyzzyva
> Date: Tue, 14 Mar 1995 07:15:21 -0600
> From: Randy Terbush <randy@zyzzyva.com>
>
> > I certainly think that these warnings should be fixed, and the coding
> > style in general needs improving. However, right now I think we want
> > to avoid changing too many source files. Not only would we move further
> > away from the NCSA source, but it also makes it a little harder to
> > integrate patches with one-another.
> >
> > The majority of your changes are to remove redundent variables (of which
> > there are sadly too many) -- which an optimising compiler would do anyway,
> > and to insert parenthesis around assignments in conditional expressions.
> > May I suggest that you trim down the patch to include only the bug fixes?
>
> Point well taken. I will trim this and resubmit it.
>
> Aw, drat. The content-negotiation stuff really does need another
> pass, and I was looking forward to finally being able to use -Wall
> while debugging it. (I suppose I could have the Makefile turn it on
> only for http_mime_db.c for the time being).
>
> Brian and Cliff have both been doing integration --- has this been a
> problem so far?

I haven't had a problem. For the record I've integrated every patch
posted as of last night around 8pm, with the exception of two of Roy's
(the log file patches, logging of referrer needs some discussion as I
think we want to see customizable logging, right? and the one using
open()/close() instead of fopen()/fclose() as it was causing core dumps -
I'll look into that later). I've done them all by hand to ensure they
didn't conflict, that was a good way to learn my way around the code
better as well. Anyways, I'll put in the patches posted today, and try
to get in the stuff I need for hotwired (DBM mods, DOCUMENT_ROOT, and
configurable default time format for http_include.c) and if Cliff doesn't
mind maybe we can see a release of apache-pre .002....(.010? :)

Brian


--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@hotwired.com brian@hyperreal.com http://www.hotwired.com/Staff/brian/
Re: Compiling with -Wall [ In reply to ]
> Date: Sat, 11 Mar 1995 21:52:51 -0600
> From: Randy Terbush <randy@zyzzyva.com>
> I made some changes to the first apache-pre release to fix some
> warnings generated when compiling on NetBSD-current with -Wall
> flag for gcc. Some of these warnings are truely anal, but it
> often does find some real bugs, and I may have found a couple
> in this case. Do with it what you want. There is also some
> patches that are specific to BSD4.4, and some specific to
> BSD which I think Brian has already caught. Would someone
> please look closely at ~185: util.c to make sure that I read
> this test correctly?
> ...
> My patch is on hyperreal in the incoming directory. (patch.wall)

I certainly think that these warnings should be fixed, and the coding
style in general needs improving. However, right now I think we want
to avoid changing too many source files. Not only would we move further
away from the NCSA source, but it also makes it a little harder to
integrate patches with one-another.

The majority of your changes are to remove redundent variables (of which
there are sadly too many) -- which an optimising compiler would do anyway,
and to insert parenthesis around assignments in conditional expressions.
May I suggest that you trim down the patch to include only the bug fixes?

David.
Re: Compiling with -Wall [ In reply to ]
On Mar 14, 9:18am, Robert S. Thau wrote:
}
} Aw, drat. The content-negotiation stuff really does need another
} pass, and I was looking forward to finally being able to use -Wall
} while debugging it. (I suppose I could have the Makefile turn it on
} only for http_mime_db.c for the time being).
}
} Brian and Cliff have both been doing integration --- has this been a
} problem so far?
}
} rst


Not so far...I've been modifying a few patches (changing some names,
adding #defines to httpd.h to allow easy tuning, etc). I have
not gotten into the high risk patch area (like content negotiation
and the shared name cache) just yet. I will send a list of which ones
I've integrated once I catch up to all the patches on hyperreal.

Cliff
Re: Compiling with -Wall [ In reply to ]
It sounds as though there aren't any actual problems with patches to tidy
up the code, so I withdraw my suggestion.