Skip to content

Conversation

@furszy
Copy link

@furszy furszy commented Aug 24, 2021

@furszy furszy added this to the 5.4.0 milestone Aug 24, 2021
@furszy furszy self-assigned this Aug 24, 2021
Making SOURCEDIST deterministic is needless since a git archive is used
as the source tarball.
Some EXTRA_DIST content is needless since a git archive is used as the
source tarball.
This commit removes the directory that is no longer used since bitcoin#16667.
@furszy furszy force-pushed the 2021_gitian_version branch from ca7a7f0 to ccd3c71 Compare September 2, 2021 22:44
@furszy
Copy link
Author

furszy commented Sep 2, 2021

rebased after 2525 merge.

@Fuzzbawls
Copy link
Collaborator

Been putting this through the gitian grinders the past days (both pre and post refactor)...and going to have to give it a NACK as it actually breaks gitian builds as-is. Further, it introduces discrepancies between CI builds and gitian builds by way of having a gitian-exclusive method of generating a source tarball, while CI builds still use make dist.

I believe that this is a prime example of a merged upstream change that was made prematurely, and without sufficient review, as Corey notates in his comments of the upstream PR post-merge. The purpose of make dist is to provide a tarball that doesn't depend on autotools whatsoever, and this breaks that by actually introducing autotools as a dependency into the source tarball build requirements...that is a glaring regression IMHO, and the upstream acceptance reads as "well, damage done...oh well", rather than correcting the issue.

@furszy
Copy link
Author

furszy commented Sep 5, 2021

About the first paragraph,
i run gitian twice with this and all went well. Weird that is failing there.
As a general point, a CI - gitian difference isn't really a reason for not adding a change (while the change is an improvement). The CI is for automated testing, need to cover and be adapted to whatever we do in build system. Not in the other way around.

About the second paragraph,
As far as i have seen, Corey didn't care much to push a PR reverting it in upstream. I'm fine closing it, makes sense 👍 .
I updated the genbuild area merely because of the "dirty" version topic. At the beginning wanted to be 1-1 with upstream, then started playing a bit with the relic-conf file ignore (which just saw you push a PR up for it, cool 👍), then got side-tracked with the Sapling related work that was doing and stopped working on this..

So.. all good for me here. Let's move forward, will give a try to 2541.

@furszy furszy closed this Sep 5, 2021
@furszy furszy deleted the 2021_gitian_version branch November 29, 2022 14:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants