Re: casper-scm Recent mlib_devel pushes to casper-astro
I think this is a good idea anyway, since the undoing bunch of commits message was to restore iBOB + BEE2. So I think it's natural to have a branch in the library. On Tue, Feb 12, 2013 at 2:35 PM, Jason Manley jasonman...@gmail.com wrote: Sorry Dave! The updates all happened when we updated all the tutorials for our recent workshop. There should have been more consultation. If 14.x is breaking backwards compatibly, maybe we should freeze mlib_devel as mlib_11.x as we did with the old v7.1 blocks when we moved to 11.x. It would then be up to 11.x users to backport/cherry-pick from the development branch. I'm afraid we don't have the capacity to maintain multiple versions and SKA-SA's firmly on the 14.x + 2012b flow. Jason On 12 Feb 2013, at 21:12, David MacMahon wrote: What's going on with recent mlib_devel pushes to casper-astro? Why so much activity from our South African friends there instead of in the ska-sa repository? Why the lack of any discussion (e.g. on this list) about this? I fetched from casper-astro on January 24th and got commit 9940104, but now I see that 18 cherry picked commits and an Undoing bunch of commits commit have been pushed to casper-astro since then. It's really hard to tell what's going on there. I dislike commit 2df7b03 (cherry picked from f22a98c) that puts 14.2 into system.xmp for the sole purpose of avoiding a rev-up when building. That may save 14.2 users a few seconds off their hours long build, but it forces every ROACH2 user to upgrade to 14.2. Do we really want to mandate that users get 14.2 if they want to do ROACH2 development? Not everyone gets donations and/or free upgrades of the toolflow software. Thanks, Dave
Re: casper-scm Recent mlib_devel pushes to casper-astro
Hi Dave, all The flurry of updates were a mistake (hence the undo). In preparing for the workshop I started seeing what updating casper-astro to ska-sa would look like and did a bunch of cherry-pick commits on my local repo. After the workshop I did a push. Realising that I had broken support for BEE2 etc I undid things. My intention was to only push bug-fixes to casper-astro, and then do a major update once a year. I have backported/merged everything from ska-sa and casper-astro (minus this latest flurry of activity) including exporting ska-sa's latest xps_library.mdl and casper_library_scopes.mdl files to Simulink 7.2 format. It also includes the new yellow block for the ADC16x250-8 board. Please take a look at the adc16-branch of this git repo: http://astro.berkeley.edu/~davidm/mlib_devel.git I think this represents the best of all worlds in that it includes all the latest pushed ska-sa fixes yet is also backwards compatible with previous tool versions. Plus it include some fixes that are unique (i.e. verilog fix for qdr controller and improved handing of newer dspsigops/Downsample versions). If people agree with me, I'd be happy to push this to casper-astro (overriding, but not eliminating, the latest stuff that's there). I think that would be great and is in keeping with making the most people happy. I still don't know what to do going forward. Support for older architectures stops at a certain point. Also, it would be nice to try to keep up with the latest versions of software while at the same time not excluding those with older versions (it really is a pain (and expensive) setting the toolflow up and making sure it works). Regards Andrew
