|
From: Julian S. <js...@ac...> - 2005-11-15 18:53:45
|
I'm just doing a pass through the docs and came across this configure-time flag. Does it do anything any more? It's still in configure.in, but I have no idea what it's for. J |
|
From: Tom H. <to...@co...> - 2005-11-15 19:05:02
|
In message <200...@ac...>
Julian Seward <js...@ac...> wrote:
> I'm just doing a pass through the docs and came across this
> configure-time flag. Does it do anything any more? It's still
> in configure.in, but I have no idea what it's for.
It cause HAVE_TLS to be defined (or not defined for --disable-tls)
without the configure test being run. I believe that define is then
used in some of the regression tests.
The problem I believe is that the configure test for TLS fails to
detect some cases correctly.
Tom
--
Tom Hughes (to...@co...)
http://www.compton.nu/
|
|
From: Nicholas N. <nj...@cs...> - 2005-11-15 19:28:38
|
On Tue, 15 Nov 2005, Julian Seward wrote: > I'm just doing a pass through the docs and came across this > configure-time flag. Does it do anything any more? It's still > in configure.in, but I have no idea what it's for. The only code that uses HAVE_TLS are none/tests/tls*.c. I don't know if they're important. none/tests/tls is one of those annoying tests that fails sporadically (as is 'yield', which I'm really tempted to just remove). N |
|
From: Nicholas N. <nj...@cs...> - 2005-11-15 19:29:44
|
On Tue, 15 Nov 2005, Tom Hughes wrote: > The problem I believe is that the configure test for TLS fails to > detect some cases correctly. IIRC Jeroen posted a patch to address this, but I don't remember the details. It was quite involved. Nick |
|
From: Jeroen N. W. <jn...@xs...> - 2005-11-15 20:24:03
|
> On Tue, 15 Nov 2005, Tom Hughes wrote: > >> The problem I believe is that the configure test for TLS fails to >> detect some cases correctly. > > IIRC Jeroen posted a patch to address this, but I don't remember the > details. It was quite involved. I did, and it was. It used code similar to none/tests/tls.c in the configure script to determine if TLS actually works. But I have seen this testcase sporadically fail in the nightly regression tests (somebody mentioned a possible race condition), so in retrospect I doubt the usefulness of my patch. Jeroen. |