Don't heavily compress packages for local builds
Categories
(Firefox Build System :: General, enhancement)
Tracking
(Not tracked)
People
(Reporter: agashlin, Unassigned)
Details
Attachments
(1 file, 1 obsolete file)
|
1.68 KB,
text/plain
|
Details |
mach package takes a long time compressing the package, even if the build is never going to leave the local machine. Maybe for local builds we can replace the args to 7z with -mx=0, and replace zip -r9D with -r0D (both of these set "store" mode, no compression).
I'm attaching a patch that does this, but not for review; it doesn't have the check for a local build as I'm not sure how to do that correctly, and not sure when I'll have time to figure it out. I've been applying this to local builds when I need to test out packaging.
Comment 1•5 years ago
|
||
I'm attaching a patch that does this, but not for review; it doesn't have the check for a local build as I'm not sure how to do that correctly, and not sure when I'll have time to figure it out. I've been applying this to local builds when I need to test out packaging.
This is a fine notion. Local builds used for performance testing are already subtle, and I expect the folks involved will already have to take steps to work around various local-only changes, so we can just add to those.
I think that DEVELOPER_OPTIONS is the simplest way to determine if we are building for local consumption, which is NOT --enable-release. With that guard I think we should flag this for review in the build peer queue.
Comment 2•5 years ago
|
||
Do you actually care about the packages? It seems to me what performance testing people want is dist/firefox/dist/Firefox.app, and you can get that with mach build stage-package (which we could add a mach package --stage "alias" for).
| Reporter | ||
Comment 3•5 years ago
|
||
Thanks, I didn't know about stage-package!
As I work on the Windows installer I do often need to build the full installer, though. Not sure about performance testing. It may be that this isn't generally useful enough to live in-tree.
| Reporter | ||
Comment 4•5 years ago
|
||
Description
•