Fixing basic Ubuntu CMake Build#1309
Conversation
|
Can one of the admins verify this patch? |
|
Thanks for your pull request. It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). 📝 Please visit https://cla.developers.google.com/ to sign. Once you've signed, please reply here (e.g.
|
|
I signed it! |
|
CLAs look good, thanks! |
|
Jenkins, test this please. Testing for core/lib/io change. |
|
|
||
| // SIZE_MAX is not defined for older gcc versions (5.1 and back). Define if | ||
| // necessary. | ||
| #ifndef SIZE_MAX |
There was a problem hiding this comment.
I wonder if this should be somewhere else, like in macros or types.h though
There was a problem hiding this comment.
Good suggestion. Moved this to macros.h
|
Our tests are down for the weekend, so we'll have to wait a bit. Sorry about that. |
|
@martinwicke No problem. I made a couple small fixes in the meanwhile. As a quick addendum, what would be the best way to get the cmake build onto the jenkins test? I found that as I was making changes, I accidentally broke the build a couple times. It would be convenient to have jenkins catch this. |
|
@rbharath can you please squash the commits? It would be great to have cmake working! I will try to set up jenkins build for cmake later this week. |
|
We found a Contributor License Agreement for you (the sender of this pull request), but were unable to find agreements for the commit author(s). If you authored these, maybe you used a different email address in the git commits than was used to sign the CLA (login here to double check)? If these were authored by someone else, then they will need to sign a CLA as well, and confirm that they're okay with these being contributed to Google. |
|
CLAs look good, thanks! |
|
@jendap Just squashed the commits. Great to hear you have time to get the cmake build up on jenkins! Please let me know if I can help. Right now the cmake build is pretty rudimentary. The tutorial compiles on Mac OSX/Ubuntu, but segfaults for some reason that needs to be fixed. We also need to add SWIG and GPU support to the cmake build. I'll try to work on this some over the next couple weeks, but would be great if any other folks are interested in helping :) |
|
Oops, looks like my squashing messed things up here. I'll fix this and update this thread once everything is sorted out. |
|
As a quick update, it appears that the cmake no longer builds tensorflow head due to recent changes in tensorflow :-(. Fixing this looks tricky, and I probably won't have time till next week or so. I'll update this thread then. |
|
Yeah, we are constantly messing with the location of header files etc. On Thu, Mar 3, 2016 at 2:04 PM Bharath Ramsundar [email protected]
|
|
I forget that cmake build. I will try to get it done by tomorrrow. |
|
There is very simple #1488. That should be enough for linux cmake cpu build. I'm have set it up to run on pull request but only when pull request touch cmake. (Bash condition That should allow to run cmake tests while not showing errors on non-cmake tests. Once we get cmake working well we can enable it for all the builds. Then we would love to create windows build as well. |
|
@rbharath can you please modify this PR to contain the intended changes only? |
|
@jendap This PR was broken by changes in tensorflow, and fixing the issue looks tricky. I don't have time to work on this for a bit (probably a couple weeks), so I'll close this PR for the time being. I'll open a new PR with fixes if somebody else doesn't fix the issue in the meanwhile. |
|
Ok, thanks! I hope somebody will help. |
More DSN updates
…cm-enhanced-fixes-210325 [r1.15-rocm-enhanced] Switching ROCm TF to use the new hipfft library (instead of rocfft)
PR #728 added basic CMake support for Tensorflow. However, that PR was only tested on MacOSX, so had a few errors on Ubuntu. This PR adds in fixes to make basic model compile on Ubuntu 12.04, with gcc 4.9.2, glibc 2.15.
The compile model does not yet run to completion (segfaults). I hope to debug that issue in a follow-on PR.