* build: derive patches upstream-head ref from script path
gclient-new-workdir.py symlinks each repo's .git/refs back to the source
checkout, so the fixed refs/patches/upstream-head was shared across all
worktrees. Parallel `e sync` runs in different worktrees clobbered each
other's upstream-head, breaking `e patches` and check-patch-diff.
Suffix the ref with an md5 of the script directory so each worktree writes
a distinct ref into the shared refs dir. Fall back to the legacy ref name
in guess_base_commit so existing checkouts keep working until next sync.
* fixup: also write legacy upstream-head ref and note it in docs
git am rewrites the index 2-3x per patch. In Chromium (~500K files,
70MB index) this dominated wall time: ~67 of 73 seconds were spent
rehashing and rewriting the index ~300 times for 150 patches.
- Add index.skipHash=true to skip recomputing the trailing SHA over
the full index on every write
- Force index v4 before am so path-prefix compression roughly halves
the on-disk index size (70MB -> 40MB)
- Disable core.fsync and gc.auto during am since a crashed apply is
just re-run from a clean reset
- Apply patch targets in parallel (capped at ncpu-2); Chromium still
dominates but this hides node/nan/etc behind it. Falls back to
sequential on roller/ branches where conflict output needs to be
readable.
- Prefix each output line with the target name so parallel output is
attributable
Measured on a 13-target config with 238 total patches: 73s -> 28s.
git format-patch honors diff.renames, which defaults to 'true' (rename
detection only). If a user has diff.renames=copies configured at the
system or global level, exported patches may encode new files as copies
of similar existing files, causing spurious diffs against patches
exported on other machines. Pin diff.renames=true to match git's
default.
* build: make patches/config.json an array of objects
This file was previously an object of patch_dir keys to repo values;
Now is an array of objects containing `patch_dir` and `repo` properties.
This makes other per-target properties (e.g. `grep`) possible.
* build: include Note metadata when exporting patches
* build: support keyword filtering in export_patches()
* build: add optional `--grep` arg to git-export-patches script
* build: update export_all_patches to understand new config file
* fixup! build: update export_all_patches to understand new config file
chore: make lint happy
* fixup! build: make patches/config.json an array of objects
chore: fix oops
* refactor: remove support for the old file format
There is more code using config.json than I thought, so the
effort-to-reward of supporting the old format is not worth it.
* build: update apply_all_patches to understand new config file
* build: update lint.js to understand new config file
* build: update patches-mtime-cache.py to understand new config file
* fixup! build: update apply_all_patches to understand new config file
fix: oops
* fixup! build: update apply_all_patches to understand new config file
fix minor syntax wart
* fixup! build: support keyword filtering in export_patches()
refactor: use idiomatic python
* refactor: warn if config.json has an invalid repo
* script: Python3 compatibility for utf8 conversion
The unicode() method has been renamed to str() in Python3,
add a wrapper around it to support running against both versions.
* script: don't require python2 for git-[import,export]-patches
The scripts work just fine with python3 too, so use the generic
python executable as the script interpreter.
Most setups don't even require or provide python 2 anymore,
so this saves one from having to install it just for the scripts.
When a patch targets a file using CRLF line endings, they need to be
retained in the patch file. Otherwise the patch will fail to apply
due to being unable to find surrounding lines with matching whitespace.
* chore: fix lint warnings
* chore: another try at python import errors
Looks like the problem is that dbus_mock.py is running as
a script but living in the `lib/` directory where it's part of a
module. Moving it up into the `script/` directory seems to
solve the issue.
In the GN build, libchromiumcontent is no longer a distinct library, but
merely a container for a set of scripts and patches. Maintaining those
patches in a separate repository is tedious and error-prone, so merge
them into the main repo.
Once this is merged and GN is the default way to build Electron, the
libchromiumcontent repository can be archived.
