electron/patches/chromium/resource_file_conflict.patch
Electron Bot d362be5cd9 chore: bump chromium to f156cb7658ec5382f5623a05841c6 (master) (#19096)
* chore: bump chromium in DEPS to 37cd06a295cf156cb7658ec5382f5623a05841c6

* update patches

* fix: replace NO_TRAFFIC_ANNOTATION_YET with MISSING_TRAFFIC_ANNOTATION

Refs: https://chromium-review.googlesource.com/c/chromium/src/+/1653256
2019-07-03 14:48:55 -07:00

79 lines
2.9 KiB
Diff

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Jeremy Apthorp <nornagon@nornagon.net>
Date: Thu, 20 Sep 2018 17:48:59 -0700
Subject: resource_file_conflict.patch
Resolve conflict between //chrome's .pak files and //electron's. The paths
that chrome code hardcodes require that we generate resources at these
paths, but GN throws errors if there are multiple targets that generate the
same files.
This is due to the hardcoded names here:
https://chromium.googlesource.com/chromium/src/+/69.0.3497.106/ui/base/resource/resource_bundle.cc#780
and here:
https://chromium.googlesource.com/chromium/src/+/69.0.3497.106/ui/base/resource/resource_bundle_mac.mm#50
This isn't needed on Mac because resource files are copied into the app bundle,
and are built in `$root_out_dir/electron_repack` (while Chromium's resources
target `$root_out_dir/repack`), but on Windows and Linux, the resource files go
directly in `$root_out_dir`, and so they conflict.
We don't actually ever generate Chromium's resource paks, but without this
patch, GN refuses to generate the ninja files:
ERROR at //tools/grit/repack.gni:35:3: Duplicate output file.
action(_repack_target_name) {
^----------------------------
Two or more targets generate the same output:
chrome_100_percent.pak
This is can often be fixed by changing one of the target names, or by
setting an output_name on one of them.
Collisions:
//chrome:packed_resources_100_percent
//electron:packed_resources_100_percent
See //tools/grit/repack.gni:35:3: Collision.
action(_repack_target_name) {
^----------------------------
Some alternatives to this patch:
1. Refactor upstream in such a way that the "chrome" pak names were
configurable, for instance by adding a method to ResourceBundle::Delegate that
LoadChromeResources would check.
2. Pass a Delegate that overrides `GetPathForResourcePack`, check for the
`chrome_{100,200}_percent.pak` filenames, and rewrite them to
`electron_{100,200}_percent.pak`.
3. Initialize the resource bundle with DO_NOT_LOAD_COMMON_RESOURCES and load
the paks ourselves.
None of these options seems like a substantial maintainability win over this patch to me (@nornagon).
diff --git a/chrome/BUILD.gn b/chrome/BUILD.gn
index 515cc0e0bad7795489e1a3c09887008f6ada2216..3c4191304b48c43999aebc8d96ebbbc01873f2d2 100644
--- a/chrome/BUILD.gn
+++ b/chrome/BUILD.gn
@@ -1709,7 +1709,7 @@ if (is_chrome_branded && !is_android) {
}
}
-if (!is_android) {
+if (!is_android && !is_electron_build) {
chrome_paks("packed_resources") {
if (is_mac) {
output_dir = "$root_gen_dir/repack"
@@ -1733,6 +1733,12 @@ if (!is_android) {
}
}
+if (is_electron_build) {
+ group("packed_resources") {
+ public_deps = [ "//electron:packed_resources" ]
+ }
+}
+
repack("browser_tests_pak") {
sources = [
"$root_gen_dir/chrome/webui_test_resources.pak",