On 04/02/11 08:29, Enea Zaffanella wrote:
Il 01/04/2011 21:58, David Fang ha scritto:
Thank you for helping to resolve this issue. I bet the reason I did not see this failure on powerpc-darwin8-gcc-4.0.1 was because I happened to test it against an older glpk (< 4.29) way back then. I can at least patch the test case to work with glpk>= 4.29 (and require said version in our TestDepends), and move on from there. Is this test also present in ppl-0.11.x? Will it require the same adjustment?
The two tests (ex1.mps and unboundedmin.mps) were adapted by Roberto on January 2010, but that changes did not get into PPL 0.10.2. There relevant commits are:
http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=97ce932e01294...
http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=commit;h=56ee86b9ccf00...
Hence, the output of PPL 0.11.x should be stable even when using the newer glpk versions.
Thanks Enea: I had completely forgotten all that. Can you please add an item to http://www.cs.unipr.it/ppl/Bugs/archive?
I am glad to know that the code produced for PPL 0.10.2 is OK.
Said that, the following is a common misconception:
On 03/29/11 21:44, David Fang wrote:
According to Jack Howarth (on this list, I believe), gcc-4.4 depends on ppl-0.10.x and is incompatible with ppl-0.11.x.
Having studied the way PPL is used in GCC 4.4 I can tell you there is absolutely no incompatibility. In other words, if you just change the configuration stuff that insists in wanting PPL 0.10.2 so that PPL 0.11.* is also accepted, there is no way a GCC 4.4 user can complain. In summary, at this stage insisting on using PPL 0.10.2 is a mistake.
Thanks for your report, David. And thanks to Alexander for testing on the PowerPC. All the best,
Roberto