Jost Berthold

Thanks @simonmar for making the point about compatibility with older formats. Dropping it means _dropping_ a feature. I don't think it is hard to keep a single header file full...

If I understand your proposal correctly, the "new workflow" is to never remove definitions from the GHC header file, by way of > remove the `#if 0` condition That makes...

On 06/03/2017 05:09 PM, Mitsutoshi Aoe wrote: > Sure, I didn't propose to > delete the comments about the Eden/Mercury/Perf events. We definitely > should keep them to avoid event...

Hi Karl, > retrofitted [1] ghc-events-0.4.4.0 with this suite and found the same failures, if I remember correctly. While this says very little about the cause of the issues, it...

Thanks @maoe for bumping this. I think the problem is inherent to the property tested by `write-merge`: Can a round-trip of writing back and re-parsing an eventlog yield the same...

> ...assembling the full eventlog for a run of a distributed program where all the workers would be writing their own eventlogs. You might think... (we did for some time...

Apologies, I did not see this issue (did not spend time on this recently, very busy). I can look into this problem if it is still relevant. Would need some...

Thanks for the report @Qqwy From the search results that you linked to, I do not see an immediate hint to what is going wrong here. Anyway, it seems as...

Confirming that this seems to have been fixed in `hlint-3.4.1` (the one which I tested). One needs to have the `LANGUAGE OverloadedRecordDot` in the file (as opposed to putting it...