Casey Rodarmor
Casey Rodarmor
I like short, inscrutable names, so I'd just call it `[defer]`, but `[defer-error]` is probably better.
Seems reasonable! Needs some tests though. Should definitely test edge cases, like an unclosed backtick, etc.
I could not wrap my head around what `split_inclusive` was doing, so I replaced it with a regex. This also avoids color codes sometimes when they aren't needed.
It looks okay for me, but that's probably just my color theme. I can kinda see how it stands out too much See below. Can you post a screenshot of...
Changed the color to cyan in #2469. Hopefully that looks good in more color schemes!
Good find! One issue though is that after the commit, the write transaction doesn't exist, so we would have to create a new write transaction, se the statistic, and commit...
I think we should just double-commit if it's the first commit. I.e., if it's the first commit and we just comitted, set the statistic and re-commit. The second commit should...
I wish we had a general solution for these kinds of thing. This is an instance of the more general problem that you cannot run a command which exports variables,...
That seems reasonable. What's the use-case? And, why not `export ANDROID_SDK_HOME := "~/Android"` in the outer scope? 'This would be slightly awkward as-is, since it would create a variable which...
Since this is really about having a per-recipe scope for expressions, then I think we should think about reusing the existing expression syntax. I.e., `FOO = "BAR"`, `export FOO :=...