Nixpkgs-update/r-ryantm logs

Thanks for the prod.
2019-09-26 logs

1 Like

2019-10-20 logs

1 Like

thanks :slight_smile: unfortunately I’m already at capacity with all the python changes recently, hopefully someone else can look at the updates :slight_smile:

1 Like

2019-11-1 logs

1 Like

2019-11-25 logs

1 Like

@ryantm you got some logs for me :slight_smile:

I use them to see if there’s any stuck python updates :slight_smile:

1 Like

2019-12-08 logs

1 Like

Merry Christmas!
2019-12-23 logs

1 Like

2019-01-14 log

1 Like

Just want to say thank you for continually doing this :slight_smile:

2 Likes

:robot: :scroll:
r-ryantm logs are now automatically available here:

https://r.ryantm.com/log/

The next step is to make the file name stay a consistent length!

Thanks to whoever has been looking at these and fixing auto updates!

Sincerely,
Ryan Mulligan

5 Likes

I kind of miss going through those and trying to fix builds. Now I feel guilty if I’m not reviewing PRs as there’s a huge backlog of them now

1 Like

Apparently it cannot update zulip because FAIL nix eval failed for "pkgs.zulip.src.urls"nix eval failed for "pkgs.zulip.urls". That’s probably because zulip uses appimageTools.wrapType2 instead of stdenv.mkDerivation.

Why does nixpkgs update need to eval pkgs.zulip.src.urls? I thought it just basically sed-replaces the old version with the new one.

@ryantm is there any chance of making the logs searchable? I’m trying to understand why tanka is not being updated, and I’m not sure if there’s a problem, if it’s just that the interval length is longer than I expect, or something else.

I noticed that the vast majority of r-ryantm update PRs are biased towards the first half of the alphabet. There are plenty of PRs for packages starting with letter A through L, but hardly any PRs for packages beyond the letter L.
@ryantm is this an unlucky coincidence, or is there a particular reason that the packages beyond L are held back from automatic updates?

2 Likes

The updates are done in alphabetical order and it might be crashing before it gets to the end… I’ll have to investigate.

There should probably be some notion of a denylist, which should avoid attrs which are “troublesome” (either crash or take forever)

There is https://github.com/ryantm/nixpkgs-update/blob/6bf8249627dea7c0dc4a6127e0fea5e553866374/src/Skiplist.hs

1 Like

impressive :slight_smile:

pretty long list, but i guess quite small in comparison to some 60k+ packages

I reversed the order of the Repology update list, so it will start at the end of the alphabet now.