tomberek

Failure seems to be in the check phase due to not having a GPU to load the relevant models. Not sure how this was handled in the past.

Looks like this is being worked in the PKGBUILD. Is there anything needed from us? Looks like a valid bug to be fixed.

I'm a fan of (B) and (C) (A): I'm looking at a use-case where we can do frame deduplication: https://github.com/facebook/zstd/pull/2737/files#diff-50e7772e96927d183da7ec7cac6a9f4e34f7e33ed00ca518611d0c52e0af4350R103 and magic CRC values would be awkward. (B): Retain the...

@yotann for (C), a slight modification would be to utilize the Seek_Table_Descriptor bitfield to indicate the presence or lack thereof of the dictionary entries. I suspect dictionary-in-stream will not be...

From meeting, we are recommending a builtins.readSymlink as a first step. Currently unsure about the interaction should be between canonicalPath and virtual paths (and related WIP changes).

I'm not clear about the implications of: > Note the subsequent if (!absolute) [...]: the certificate and key are not configured for such "absolute" requests, as their URL components would...

A narHash alone is not quite enough to avoid a download. There are more attributes needed that are being provided by `~/.cache` and thus forces a reload when those are...

Some of the attributes like ref/rev are to allow reliable fetching. The local short-circuit with just narHash is helpful, but doesnt help fetch the correct thing when others are using...

Ideas to consider - current behavior is for `nix log` to interpret no argument as the "default installable", the current directory flake. So in many cases this will accidentally be...

It would take a build2 call, similar to https://github.com/NixOS/nix/blob/master/src/nix/profile.cc#L371. Notice that rumming "build" on that installable does work. And yes, templates were not intended to require a build but rather...