I'm starting to get mad Sisyphus vibes

[sly rain]

What's an example of something that's blocking you where it's not obvious?

[finite jetty]

honestly it's just the sheer number of conflicts, for example here's one single function: https://gist.github.com/alexcb/c2bba1570bbd9816254a0b2b1396ceaa

[sly rain]

That seems like an odd conflict to have to address... what commit are you on from egraph and what are you cherry-picking?

>>>>>>> 147b0b45b (workspace: plumbing & compat (#11995))

I'm confused why that would be showing up at all, for instance

[finite jetty]

currently f358ebb5401acd2203cd114fead28e889201592b is what sipsma/egraph is pointing to, which is showing up as HEAD

then I squashed all 8 commits from lockfile (pushed up to my lockfile-squash branch, 147b0b45bdad8fa9d4c6871b58bb77590a468ac1), then have been trying to cherry-pick all of that

[sly rain]

Ah, the lockfile branch is 1 week behind main

So your conflicts are probably rehashing the conflicts I already resolved between egraph and main

[finite jetty]

ah haaaa, yeah I should try tackling that seperately then.

[sly rain]

yeah I would rebase/merge from main the lockfile branch first

then merge/cherry-pick it into egraph should be much simpler

[patent cloud]

depending on strategies git could still be dumb in certain cases :S (even if doing "main first then egraph" but def worth a try)

[finite jetty]

I also had a brief attempt at generating a patch, and applying that directly, but also started getting lots of rejects.

[finite jetty]

so far it's generating much less conflicts, in fact now I'm just dealing with various renames like buildkit.ExecutionMetadata to engineutil.ExecutionMetadata

[sly rain]

Good yeah from glancing through the lockfile PR it really seems like only a couple sections would have conflicts

Please ask questions when anything comes up though, really want to start help getting people up to speed on the changes since I'm OOO starting next week for 2.5 weeks

[finite jetty]

any chance you ran into

1274: ┆ ┆ [2.5s] | github.com/gogo/protobuf/gogoproto/gogo.proto: File not found.
1274: ┆ ┆ [2.5s] | checksum.proto:5:1: Import "github.com/gogo/protobuf/gogoproto/gogo.proto" was not found or had errors.
1274: ┆ ┆ [2.5s] | engine/contenthash/generate.go:3: running "protoc": exit status 1
```?

[sly rain]

No, where did you hit that?

[finite jetty]

when I ran dagger --progress=plain generate -y under my merged branch (I ended up trying a merge since it was easier than a rebase)

[sly rain]

seeing if I can repro, but in mean time I would suspect an include/exclude filter somewhere might need updating

[finite jetty]

I was able to reproduce it using the generate command on your branch too

[sly rain]

Yeah I hit it too, weird cause I in the last week or so definitely ran dagger call generate layer export --path . successfully, something must have nudged out of place

looking

[finite jetty]

maybe we need to add another install under https://github.com/alexcb/dagger/blob/aff8498b9dd2ec5967f9971e1be8e2f7e27eb78a/toolchains/engine-dev/main.go#L400 ?

[sly rain]

This fixes it:

diff --git a/engine/contenthash/generate.go b/engine/contenthash/generate.go
index 9e465f653..e10a43ed7 100644
--- a/engine/contenthash/generate.go
+++ b/engine/contenthash/generate.go
@@ -1,3 +1,3 @@
 package contenthash
 
-//go:generate protoc -I=. -I=../../../../ --gogofaster_out=. checksum.proto
+//go:generate protoc -I=. -I=../../ --gogofaster_out=. checksum.proto

That package got migrated out of internal/buildkit recently, forgot to update the ../ ladder

[finite jetty]

yeah that makes total sense, thanks!