This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH 1/2] Bump CLooG.org version to 0.16.0.
On Wed, Dec 29, 2010 at 12:30:54PM -0500, Tobias Grosser wrote:
> On 12/29/2010 12:12 PM, Jack Howarth wrote:
>> On Wed, Dec 29, 2010 at 05:36:23PM +0100, Sven Verdoolaege wrote:
>>> On Wed, Dec 29, 2010 at 11:23:39AM -0500, Jack Howarth wrote:
>>>> I was able to build cloog-parma and it passed its testsuite, but the cloog-parma
>>>> pre-release doesn't seem to install the stride.h header from the cloog-core subdirectory.
>>>> Jack
>>>
>>> Try this:
>>> [..]
>> Sven,
>> Thanks. This eliminates the problem of uninstalled headers. Unfortunately, cloog-parma still
>> confuses configure in gcc trunk though and fails as...
>>
>> [..]
>> Jack
>> ps It appears that the version checks for cloog in the graphite branch are signficantly
>> different from those in gcc trunk so that today's cloog.org version bump for the graphite
>> branch can't be applied to trunk to solve this issue.
>
> Hi Jack,
>
> where do you see this huge difference? For me the config/cloog.m4 files
> are identical and the section in configure.ac looks in both files like
> this:
>
> if test "x$with_cloog" != "xno"; then
> dnl Version check for CLooG-Org
> dnl As long as there is no new release of CLooG,
> dnl we will check for 0.14.0.
> dnl
> dnl The first git revision that will work with
> dnl GCC is: bd91b845a65805c290d43fc1bef8139864a163fb
> dnl This is enforced implictly, as this is the commit that
> dnl introduced the versioning information used within our
> dnl checks.
> dnl
> dnl If we're using CLooG-Legacy, the provided version information
> dnl will be ignored.
> CLOOG_CHECK_VERSION(0,14,0)
>
> Cheers
> Tobias
Tobias,
My mistake. They are similar but I am confused about the sections that were left
unchanged...
--- a/configure
+++ b/configure
@@ -5971,8 +5971,8 @@ $as_echo "$gcc_cv_cloog_type" >&6; }
LDFLAGS="${_cloog_saved_LDFLAGS} ${clooglibs} ${ppllibs}"
if test "${cloog_org}" = yes ; then
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking for version 0.14.0 of CLooG" >&5
-$as_echo_n "checking for version 0.14.0 of CLooG... " >&6; }
+ { $as_echo "$as_me:${as_lineno-$LINENO}: checking for version 0.16.0 of CLooG" >&5
+$as_echo_n "checking for version 0.16.0 of CLooG... " >&6; }
if test "${gcc_cv_cloog_ct_0_14_0+set}" = set; then :
$as_echo_n "(cached) " >&6
else
@@ -5983,7 +5983,7 @@ int
main ()
{
#if CLOOG_VERSION_MAJOR != 0 \
- || CLOOG_VERSION_MINOR != 14 \
+ || CLOOG_VERSION_MINOR != 16 \
|| CLOOG_VERSION_REVISION < 0
choke me
#endif
Shouldn't these tests be using gcc_cv_cloog_ct_0_16_0+set now?
Jack