News for hackers. Stuff that matters.

naholyr

Ah ah nope, and that arised fast (issue #4) ;)

But definitely, if you find it useless, then it certainly is for you! I was just telling why others could find it useful :)

It's all about not forgetting to create the issue: you just put your // TODO and not go out of your editor.

An IDE plugin could do the trick nicely too (create an issue by right-clicking on a TODO for example). But a git hook is more portable.

In fact the first prototype was a shell script using grep and ghi (https://github.com/stephencelis/ghi).

Anyway, it depends on your workflow, if you already have the right tools and habits no to lose your TODOs in your deep code, Github-Todos is obviously of no use for you!

Hope it will do the job :) still in active development so… https://github.com/naholyr/github-todos/issues ;)
That's more the direction I think about yes
Jira should quickly be on its way, in the meantime I'll make everything needed to make the operation of adding a new issue manager a very easy task.
This coupling is not ready yet, I still have to find a satisfying workflow.

When issue is closed :

— it detects TODOs anywhere, not necessarily in comments, it could destroy some code (same for 'inject-issue' but here it would be a removal, can be tougher) — maybe a TODO previously related to this issue will have to be modified instead of deleted, how should it handle it?

When TODO is removed :

— should I close issue directly or check in the whole codebase if it still exists somewhere and keep it open?

Many questions arise and I still found no harmless answers :(

But it definitely should help you track it both ways, with a command "stat" for example that would report you any actions you should achieve, and even interactively do it for you. But I don't think it will happen in hook itself yet.

I want all ideas available here ;) should create an issue for open discussion.

Hi I'm the author :)

Note that it can open or comment issues (matching based on TODO's text == issue's title), but it does not yet closes them, because I couldn't find a satisfying workflow yet (which means it may never occur).

Give it a try on a sample repository, always needing more testers !

In this situation, "foo" or "bar" or both should not depend on baz, but have it as a peer dependency. There is nothing broken, just a very specific case not so easily identifiable, but very easily fixable.