Matthew Pickering to

A very true and important tweet. It took me a long time to come to terms with the trauma I have faced and learn to deal with it in a healthier way.

@ChShersh @angerman Are you suggesting that we shouldn't prepare minor releases which fix bugs? It also seems inaccurate to suggest that we don't spend a significant amount of time ensuring that new releases do not introduce unexpected regressions when compiling packages.

@angerman If only all this time spent on backporting features and bug fixes was instead spent on making new versions to be used easier…

@angerman What is the ticket for the regression? The release passed the testsuite on all supported platforms and was tested for regressions by compiling stackage. If there is missing coverage then it needs to be integrated into the CI pipeline.

@domenkozar gitlab.haskell.org/bgamari/ghcs-n…

Would anyone benefit if we created nixpkgs-ghc, where every single version of GHC ever released would be available for the 4 common platforms?

@csaba_hruska Looks like a catch-all case in case there is anything which isn't supported.

@mpickering_ What is this? #L370" target="_blank" rel="nofollow noopener">gitlab.haskell.org/ghc/ghc-debug/…

AFAIK ghc-debug can not decode all closure types. (right?) So I plan to make an experiment manually, I'd like to modify the GHC RTS GC to dump the whole heap to a file during the traversal and then analyse it with custom offline tools.

@csaba_hruska No, you can use it for offline analysis.

@mpickering_ Great! Does it contain enough info to rebuild the native program state from scratch and resume execution?

New blog post by @mpickering_: Interface Files with Core Definitions. Speed up Template Haskell compilation times by up to 40%! well-typed.com/blog/2023/02/i…

This completely arcane GHC RTS bytecode interpreter segfault is easily the most fun bug hunt I have gone on in quite some time #note_482997" target="_blank" rel="nofollow noopener">gitlab.haskell.org/ghc/ghc/-/issu…

@fellrunningguid In the cloud all day over towards Bleaklow!

Worth getting up early for my run today #peakdistrict #fellrunning #trailrunning

@fellrunningguid @harveymaps Have to go straight, terrain is all very runnable on the plateau with big catching feature. North from hill to catch the big grough heading E, then bearing from junction.. hoping to catch the only grough running E but plenty of options to catch inaccuracy.

Navigation Task: Route choice - Get from A to B in bad visibility Timing, pacing, attack points, tick offs, catching features? Also consider terrain. Options explained for Patreon subscribers this evening. @harveymaps Superwalker used with licensing agreement #mapreading

Don't forget to join our tutorial on ghc-debug and eventlog2html on Thursday 1 December at 1900 UTC: well-typed.com/blog/2022/11/l…

@FinleyMcIlwaine will give a workshop at Zurihac tomorrow about using ghc-debug to find realistic and difficult to find space leaks - #FinleyMcIlwaine" target="_blank" rel="nofollow noopener">munihac.de/2022.html#Finl…

Yesterday I released the 0.3 series of the ghc-debug packages, which includes significant improvements to the reliability and interface of the TUI. Many thanks to @FinleyMcIlwaine, @kosmikus and Teo for their contributions.

The Deep Subsumption proposal invites the community to comment on a technical solution to the issues introduced by the simplified subsumption proposal. github.com/ghc-proposals/…

The current status of the 9.0.*, 9.2.* and 9.4.* release series: haskell.org/ghc/blog/20220…

If you are thinking of attending ZuriHac but finances are holding you back, we have some scholarships available!🎓🏔 (RT for reach very much appreciated) #scholarships" target="_blank" rel="nofollow noopener">zfoh.ch/zurihac2022/#s…

@istathar @donsbot Perhaps the structure of your module graph doesn't allow a high degree of pararrelism. The model is very simple.

@acid2 You should try using the TUI, happy to show you how on a call if you like.

@mpickering_ Schweeeet! I am absolutely craving it today. I have an OpenCV image that I can see is retained (as determined by a dyepack-like WeakPtr), but I have no idea why. I think ghc-debug would be a great fit to find out what's keeping it alive

@mpickering_ What's the status of ghc-debug? Does it still need a custom GHC?