Comments: Raindrops from a brainstorm

Two things:

1) Brendan implemented line number annotations that the preprocessor should be outputting so that exception line numbers match the source file... except then view-source to see the error is broken. It's not clear what to do about this.

2) Requiring any sort of large patch to merge cleanly to be reviewable means it'll never be reviewable. Those rot on a weekly basis for sure, some more often.

(From Alex: Hmm. Good point.)

Posted by Boris at July 14, 2007 7:08 PM

The Bugzilla part sounds very useful to me. Making distributed patch handling optional and/or automatically turning it off based on the patch size and/or number of touched files should already be enough to cover cases like the one brought up by Boris.

Posted by Dao at July 15, 2007 1:39 AM

JDeveloper 1.3 has an in-built change history that lets you restore or delete snippets of changes that you've made.

I don't know if it's powerful enough (or even appropriate) for your needs, but I suspect that your experimental change sets problem could be and may already be solved by such an implementation. The obvious downside is that you'd be tied to specific IDEs if you depend on the feature.

Posted by Ben Basson at July 15, 2007 3:25 AM

Doesn't the preprocessor already emit source lineno annotations?

(From Alex: Yes, it does, but it's not enough, not nearly enough. I want a tool that will read those annotations, let me edit in one file, and reflect the changes in the matching file.)

Posted by Benjamin Smedberg at July 15, 2007 10:12 AM