cURL / Mailing Lists / curl-library / Single Mail

curl-library

Re: [autobuild] those FreeBSD warnings

From: Daniel Stenberg <daniel-curl_at_haxx.se>
Date: Wed, 3 Mar 2004 11:02:10 +0100 (CET)

On Tue, 2 Mar 2004, Joe Halpin wrote:

> I emailed the maintainer of the FreeBSD curl port, but never got a response,
> so I'm assuming he doesn't mind sharing his patches.

I'd rather get specific permission from the person who actually wrote patches,
than just assuming he wants us to use them.

I'll also illustrate below why this secret-patching-without-telling-us is a
bad practise. They're not all very good.

> I'll attach them here, but I'm not sure we need them, since the build
> looks ok now (although we might want to consider patch-lib::url.c).

(Thanks for posting these patches.)

I had a look at the patches, and this is my verdict:

1. "patch-configure" - prevents configure to add extra -I flags to the path
   if it is in "/usr". A reasonable fix.

2. "patch-lib::hostip.c" - a weird bug fix, which has been corrected in a
   better way in curl's CVS

3. "patch-lib::url.c" - a very dubious fix which I *really* would like to
   hear an explanation for. It effectively prevents curl from timing out
   name lookups by default.

4. "patch-ltmain.sh" - I can't comment on this since I haven't really
   figured out what the patch disables. But, we've upgraded libtool since then
   so possibly this patch isn't needed anymore. This patch patches code
   generated by a 3rd-party tool.

5. "patch-src::main.c" - completely wrong patch. The libcurl interface takes
   a long pointer, even when representing time.

6. "patch-tests::server::getpart.c" - unnecessary patch. The single if()
   should be enough to extend the allocated memory to the proper size, without
   looping.

> I must have fat-fingered something earlier.

No I don't think so, I fiddled with the configure script so that it won't use
that option with your version of gcc, since it can't hide warnings on system
headers properly.

I want 'configure --enable-debug' to set as aggressively pedantic options as
possible, but without causing "false positives".

> There still are a couple warnings about select() not being prototyped in the
> tests/lib files. I'll send patches for that separately.

Thanks, I've read that mail and I'll make the required changes.

-- 
    Daniel Stenberg -- http://curl.haxx.se/ -- http://daniel.haxx.se/
   [[ Do not send mails to this email address. They won't reach me. ]]
Received on 2004-03-03