alexjski

My current thinking is that I will try as hard as I can to remove that mostly to remove complexity. That is not to say that this is a done...

> Well, that might be easy to answer. The reason I added `BAZEL_TRACK_SOURCE_DIRECTORIES` back then was because `glob` converts filenames into labels, and there are file names that aren't valid...

> Well, that might be easy to answer. The reason I added `BAZEL_TRACK_SOURCE_DIRECTORIES` back then was because `glob` converts filenames into labels, and there are file names that aren't valid...

> Do you have any expectations regarding the next steps? Are there any obvious issues with source directory tracking that would need to be resolved or is it more of...

> @alexjski Friendly ping. I resolved the merge conflict. Sorry, it fell off my radar. Looking pretty good, added only some minor comments.

I am afraid that this is a way to pop up internally now (in fact, quite a lot; to make matters worse, in that case it is benign), so if...

I have some questions and comments here: 1. The presence of the empty directory in the expansion may be a little more complicated API. We include directories in there, but...

I wonder if at this point it would not be productive to draft a doc to discuss the problem and alternatives considered (https://bazel.build/contribute/design-documents). In particular, I would expect that most...

> @fmeum I tested your commit that stages it as a symlink pointing to the real symlink (option 1 above), [fmeum@8e9fa6a](https://github.com/fmeum/bazel/commit/8e9fa6a45c0735f5d433e1118455de61a64bff12). This fixes the sandbox issues with symlinks in rules_js....

> @alexjski Can you give an example? You'd only get a symlink to a symlink in the sandbox if you have the symlink outside the sandbox as an input. >...