| programmerjake | basically just push your changes to master after testing them | 00:06 |
|---|---|---|
| programmerjake | lkcl, we should start thinking about maybe changing that system since it only works if there aren't many people contributing...we're no longer just 3-4 people | 00:08 |
| Las[m] | You could optionally try making use of Gitea or similar to have GH-style PRs | 10:36 |
| Las[m] | Or you could just use e-mails | 10:36 |
| ghostmansd | Hi folks, I'm back from sick leave, it took longer than expected. I'll be able to start where I left on weekends, though. | 14:17 |
| lkcl | ghostmansd, star | 14:17 |
| kylel | lkcl, the state class is broke on your commit. there's no depth attribute on the sim side (at least that I could find...which is why I walked the list) | 15:15 |
| lkcl | kylel, argh, my fault, i walked away from the laptop last night without committing the fix after discovering that | 15:44 |
| lkcl | ok done, pushed | 15:46 |
| lkcl | what i meant was, call the new get_mem() *and* call the *existing* (old, soon-to-be-replaced) check_sim_memory() function | 15:47 |
| lkcl | temporarily | 15:47 |
| lkcl | then, deliberately break ISACaller (by temporarily editing decoder/isa/fixedload.py to produce totally wrong data) | 15:48 |
| lkcl | and check that they both find the error | 15:48 |
| lkcl | really we need proper unit tests for TestState to make sure it actually detects borked memory (and regs) | 15:48 |
| lkcl | i'm not entirely sure how to test that *unit tests* actually raise exceptions! | 15:49 |
| kylel | gotcha | 15:49 |
| kylel | my method is to undo my actions that bork them :) | 15:50 |
| kylel | Need assertfalse | 15:50 |
| lkcl | ah ha! | 15:50 |
| lkcl | no it's very simple | 15:50 |
| lkcl | @unittest.expectedFailure | 15:50 |
| lkcl | https://www.kite.com/python/examples/1299/unittest-add-an-expected-failure-test | 15:51 |
| lkcl | because we don't want to have to develop one set of code which has assertFalse *instead* of assertTrue | 15:51 |
| kylel | nice and simple | 15:57 |
| lkcl | yes. and makes sure that the *actual* unit test code can be tested [for expected failures] without TestState needing to be modified. | 16:39 |
| lkcl | i'd seen that trick somewhere. | 16:40 |
| programmerjake | Las: the last time we discussed using gitea: https://bugs.libre-soc.org/show_bug.cgi?id=131 | 18:18 |
| Las[m] | programmerjake: Thanks for the link, not sure I can help set up Gitea if it's not NixOS. | 18:31 |
| Las[m] | I think what I'll do is make a separate branch somewhere, which one of you can review and merge if you want | 18:32 |
| Las[m] | For comments, etc., we could just use this channel | 18:32 |
| kylel | lkcl, check_mem checks out. | 18:50 |
| lkcl | ack, excellent | 20:00 |
| lkcl | Las[m], the time and cost of anything like gitea is too great. | 20:00 |
| lkcl | gitolite3 combined with gitweb is bare-bones and minimal attack surface | 20:01 |
| Las[m] | I don't think the attack surface is too important if you sandbox it and sign your commits | 20:01 |
| Las[m] | And manually push commits from some Gitea-controlled repository to some public mirror | 20:02 |
| programmerjake | also, gitea is quite lightweight...nothing like gitlab | 21:49 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!