Hi friends,
There are two test programs that do not compile if the platform does not
support strdup(), even if HAVE_STRDUP is undefined: lib509.c and
lib1900.c.
Does anybody has some hint how to make them compilable, still preserving
them for platforms implementing strdup() ?
The library
Does anybody has some hint how to make them compilable, still preserving
them for platforms implementing strdup() ?
Attached (completely untested) patch should fix the issue. At least,
as long as libcurl isn't built as a shared library with symbol hiding
enabled.
Give it a try,
--
-=[Yang
On Jul 12, 2013, at 5:58 PM, Yang Tse yangs...@gmail.com wrote:
Does anybody has some hint how to make them compilable, still preserving
them for platforms implementing strdup() ?
Attached (completely untested) patch should fix the issue. At least,
as long as libcurl isn't built as a shared
On Fri, Jul 12, 2013 at 12:12 PM, Ryan meok...@yahoo.com wrote:
On Jul 12, 2013, at 5:58 PM, Yang Tse yangs...@gmail.com wrote:
Does anybody has some hint how to make them compilable, still preserving
them for platforms implementing strdup() ?
Attached (completely untested) patch should fix
Yang Tse wrote:
Attached (completely untested) patch should fix the issue. At least,
as long as libcurl isn't built as a shared library with symbol hiding
enabled.
Give it a try,
Your patch is OK for lib1900.c, but not for lib509.c because it uses
(strdup)(ptr)
However by changing your
implementing strdup() ?
Attached (completely untested) patch should fix the issue. At least,
as long as libcurl isn't built as a shared library with symbol hiding
enabled.
Yeah, don't you think it would be a good idea to test the patch first before
anyone does anything?
Patrick is one
On 12 July 2013 15:18, Ryan meok...@yahoo.com wrote:
Upon further examination, I see that the patch is quite trivial and not worth
any fuss. But to an initial observer, saying you're submitting an untested
patch sounds a bit sketchy to me. You obviously have a different opinion.
I hope you
On Jul 12, 2013, at 11:34 PM, Tor Arntsen kspt@gmail.com wrote:
On 12 July 2013 15:18, Ryan meok...@yahoo.com wrote:
Upon further examination, I see that the patch is quite trivial and not
worth any fuss. But to an initial observer, saying you're submitting an
untested patch sounds a
Patrick Monnerat wrote:
Your patch is OK for lib1900.c, but not for lib509.c because it uses
(strdup)(ptr)
However by changing your patch to
#define strdup curlx_strdup
It compiles fine with that.
Hi :-)
I don't foresee any problem in libtest programs if your #define was
used
On Fri, Jul 12, 2013 at 07:57:04PM +0200, Yang Tse wrote:
If memory serves me, long time ago Dan Fandrich also had some
platforms without strdup.
That does sound familiar, but I lost access to my mini-build farm of odd and
obsolete platforms some time ago, and I don't recall which one
to confess that when I first looked at this, I was confused why
it was safe to check newurl in the else case. It doesn't get
initialized when it's declared. My first thought was that we could be
checking garbage on the stack. It wasn't until I realized that the
strdup in Curl_retry_request modifies newurl
11 matches
Mail list logo
