Crash when using non-empty input-renaming-map about closure-stylesheets

[elipoultorak]

That makes more sense, but it doesn't explain why the code works. :-)

[sgammon]

i have a fork of closure-stylesheets that gathers most, if not all, of these fixes and features introduced via PRs. i'll merge this in today, feel free to ping me if our fork can be helpful to you in the meantime before this gets merged to master upstream.

[elipoultorak]

Any reason this wasn't merged yet? The problem still exists, and @Yannic's commit fixes it.

[sgammon]

@elipoultorak it does fix it but if memory serves, it just disables the input renaming map? or did it work for you? if so then yeah why isn't this merged lol

[Yannic]

TBH, I don't remember what problem I was trying to solve when I initially run into this crash, and I haven't seen it for the last two years.
IIRC, #125 worked but wasn't merged because the "fix" just hides the issue that the map isn't initialized where that's supposed to happen.

[elipoultorak]

I just tried it, and for some reason, it does rename it properly with the fix. I used a CLOSURE_COMPILED format instead of json, so I don't know if it would work with json. But when I use the master branch, or the last release, it gives me an error. When I use @Yannic's fork, it uses the correct renaming according to the map.
I know it sounds weird, do you know why it happens? Maybe the SubstitutionMap is initialised more than once, sometimes using a correct class, and sometimes not?

[elipoultorak]

Update: it also works with json maps.

[Yannic]

AFAIK, the crash above happens because the map isn't initialized at all. You're right that my fix works, but the map is only initialized right before the crash would happen and not where the initialization is supposed to happen.
Unfortunately, that's all I remember right now. If I remember to do so, I'll have a look (next week or so) if I can find some notes from two years ago.

[sgammon]

@elipoultorak @Yannic sick, ha ha
another one bites the dust, good work @Yannic

[elipoultorak]

Looks good, but is nobody at Google merging anything to this repo?

[sgammon]

@elipoultorak welcome to the gap between regular open source, and google internal :) lol