Opened 12 years ago

Closed 12 years ago

Last modified 12 years ago

#35231 closed update (fixed)

cgal: update to 4.0.2 and fix build with Boost 1.50.0

Reported by: raphael-st (Raphael Straub) Owned by: Veence (Vincent)
Priority: Normal Milestone:
Component: ports Version:
Keywords: haspatch Cc: philipp.moeller@…
Port: cgal

Description

The attached patch updates cgal to version 4.0.2. It also adds a dependency on eigen3, which now can be used with cgal, and metis, which is used by some examples.

The other patch fixes a build problem with Boost 1.50.0.

Attachments (2)

Portfile.diff (1.5 KB) - added by raphael-st (Raphael Straub) 12 years ago.
patch-CGAL_SetupBoost.cmake.diff (598 bytes) - added by raphael-st (Raphael Straub) 12 years ago.

Download all attachments as: .zip

Change History (13)

Changed 12 years ago by raphael-st (Raphael Straub)

Attachment: Portfile.diff added

Changed 12 years ago by raphael-st (Raphael Straub)

comment:1 Changed 12 years ago by Veence (Vincent)

Does it compile allright? I tried 4.0.2, and it failed miserably: the Makefiles produced by CMake are wrong, no library is build… I filed a bug two weeks ago upstreams, I'm still waiting for an answer…

comment:2 in reply to:  1 Changed 12 years ago by raphael-st (Raphael Straub)

Replying to vince@…:

Does it compile allright? I tried 4.0.2, and it failed miserably: the Makefiles produced by CMake are wrong, no library is build… I filed a bug two weeks ago upstreams, I'm still waiting for an answer…

Do you mean 4.0.1, which had a build problem? 4.0.2 compiles successfully on Snow Leopard (i386) and Leopard (i386, ppc), here. I cannot test it on (Mountain) Lion.

comment:3 Changed 12 years ago by Veence (Vincent)

Yes and no. It is the same phenomenon as 4.0.1, but it plagues also the 4.0.2 for me, albeit the release manager pretends it is correct.

comment:4 Changed 12 years ago by ryandesign (Ryan Carsten Schmidt)

Cc: philipp.moeller@… added

Has duplicate #35351.

comment:5 Changed 12 years ago by Veence (Vincent)

Resolution: fixed
Status: newclosed

Done tonight in r96188. Seems to work, though not with clang, which is a bummer. We should maybe inform upstreams of this. Thanks for the patch.

comment:6 Changed 12 years ago by philipp.moeller@…

4.0.2 had multiple issues with clang. That has been fixed for the next release, which should be due soon. Until then gcc is the only viable option on MacOS. Sorry.

comment:7 Changed 12 years ago by Veence (Vincent)

No worry. It’s okay since llvm-gcc is still part of Xcode 4.4, but I think it’ll be removed next year. Anyway, it’s great news that the next release will be clang compatible. Thanks for the effort!

comment:8 Changed 12 years ago by Veence (Vincent)

Oh, BTW did you perform performance tests clang vs gcc? I think it’d be interesting.

comment:9 Changed 12 years ago by philipp.moeller@…

No. I would like to see a benchmark that strains interval arithmetic do see the actual differences between -frounding-math on gcc and the fix we have for clang. This would also be a nice indication if it is worth the effort to implement -frounding-math for clang. Feel free to benchmark something and show it. I would like to see it.

comment:10 Changed 12 years ago by Veence (Vincent)

Do you have any idea on which algorithm in the library would seem the most fit for such a benchmark?

comment:11 Changed 12 years ago by philipp.moeller@…

I'd probably try a Delaunay_triangulation_2 http://www.cgal.org/Manual/latest/doc_html/cgal_manual/Triangulation_2/Chapter_main.html#Subsection_37.5.2 Of course minus the input stream and some buffering and repeated runs and whatnot. That should be a good start but I wouldn't use the results it to predict a mars landing or anything else.

Note: See TracTickets for help on using tickets.