percona-server_8_0: fix compiling on darwin #401322
Conversation
github-actions
bot
added
10.rebuild-darwin: 1
nix-owners
bot
requested review from
Ma27,
ctheune,
dpausp,
frlan,
leona-ya and
osnyx
|
There was a problem hiding this comment.
I'm not sure why the patch includes a license. How about using fetchpatch instead?
There was a problem hiding this comment.
sorry it took me a while to get back to this!
this patch is adapted from a patch to mysql80 in nixpkgs[1] which a change in mysql 9.2 backported to 8.0[2][3], hence the license. there's an extra bit in there to apply the changes to range_optimizer/index_range_scan_plan.cc. i could split this up so i instead reference the patch out of mysql80 and then only add the range_optimizer/index_range_scan_plan.cc bit as a separate patch to percona. i just couldn't find any other examples of packages using patches from other packages and it felt better to keep things together since it's one logical change. no strong preference on my end though.
[1] https://github.com/NixOS/nixpkgs/blob/33949e56e5d67374ca143da77bcc8ab994c4a2e6/pkgs/servers/sql/mysql/libcpp-fixes.patch
[2] #374591 (comment)
[3] mysql/mysql-server@3a51d7f
There was a problem hiding this comment.
I do not think we have to split up the patch into the smallest possible parts just by principle, this can also affect readability. So I think your approach is feasible.
Another reason for splitting the patch into the base patch you adapted, and the changes you made yourself on top, could be making it easier to pull in future changes from mysql as well. But instead you can just optionally reference the base patch by URI in the comment on top of the patch that currently only holds the license.
One minor license nitpick is that, when distributing modified cersions of GPL source code, you need to clearly state that you've modified the source (2.a), adding your own copyright notice there.
Currently percona-server_8_0 fails to compile on Darwin for a few reasons. First it hits an error due to adding multiple copies of OpenSSL[1] which Homebrew has fixed and attempted to upstream[2][3]. After addressing that, it fails to compile due to the removal of std::char_traits<uchar> in LLVM 19. This has already been patched in mysql80[4], but I couldn't find a precedent for grabbing patches from another package, so I opted to adapt the patch instead. The Percona-specific changes are minimal and could be extracted into a separate patch if that's desirable. [1] https://perconadev.atlassian.net/jira/software/c/projects/PS/issues/PS-9641 [2] percona/percona-server#5537 [3] Homebrew/homebrew-core#204799 [4] NixOS#374591
joshheinrichs-shopify
force-pushed
the
percona-server-8_0-darwin
branch
from
8faa4e3 to
321d3db
Compare
yzx9
added
the
0.kind: ZHF Fixes
|
Successfully created backport PR for |
|
Git push to origin failed for release-25.05 with exitcode 1 |
github-actions
bot
added
the
8.has: port to stable
4 participants
Currently percona-server_8_0 fails to compile on Darwin for a few reasons. First it hits an error due to adding multiple copies of OpenSSL[1] which Homebrew has fixed and attempted to upstream[2][3]. After addressing that, it fails to compile due to the removal of
std::char_traits<uchar>in LLVM 19. This has already been patched in mysql80[4], but I couldn't find a precedent for grabbing patches from another package, so I opted to adapt the patch instead. The Percona-specific changes are minimal and could be extracted into a separate patch if that's desirable.[1] https://perconadev.atlassian.net/jira/software/c/projects/PS/issues/PS-9641
[2] percona/percona-server#5537
[3] Homebrew/homebrew-core#204799
[4] #374591
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.