Closed Bug 767477 Opened 13 years ago Closed 13 years ago

Add package tests and symbols for win8 builds

Categories

(Release Engineering :: General, defect, P2)

x86
macOS
defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: armenzg, Assigned: armenzg)

References

Details

Attachments

(2 files, 1 obsolete file)

Let's add such functionality as well. Right now symbols are failing but we haven't figure out yet why.
Assignee: nobody → armenzg
Priority: -- → P2
It might need MOZ_SYMBOLS_EXTRA_BUILDID as well.
Attachment #638878 - Attachment description: [wip] add packaged test and symbols for win64 → add packaged test and symbols for win64
Attachment #638878 - Flags: review?(nrthomas)
Attachment #638878 - Flags: review?(nrthomas) → review?(coop)
Attachment #638878 - Flags: review?(coop) → review+
I just realized the other patch included symbols info which is not what I wanted.
Attachment #640225 - Flags: review?(coop)
Do I need any appending like on armv6?
Attachment #640227 - Flags: review?(nrthomas)
Attachment #640225 - Flags: review?(coop) → review+
Comment on attachment 640225 [details] [diff] [review] win32-metro packaged tests 9cee0654a426
Attachment #640225 - Flags: checked-in+
Attachment #638878 - Attachment is obsolete: true
Comment on attachment 640227 [details] [diff] [review] add symbols for win32-metro (In reply to Armen Zambrano G. [:armenzg] - Release Engineer from comment #5) > Do I need any appending like on armv6? This patch looks fine for dep builds (the env changes won't be used yet). For nightlies you will probably need a MOZ_SYMBOLS_EXTRA_BUILDID defined to avoid the regular win32 desktop symbols. Once this patch gives you dep builds producing crashreporter-symbols.zip you can look inside one and see how the .txt file is named. The existing symbol manifests look like this firefox-16.0a1-WINNT-<buildid>-x86_64-win64-elm-symbols.txt firefox-16.0a1-WINNT-<buildid>-x86-elm-symbols.txt
Attachment #640227 - Flags: review?(nrthomas) → review+
This landed in production yesterday, sorry for the delay in notification.
(In reply to Aki Sasaki [:aki] from comment #8) > This landed in production yesterday, sorry for the delay in notification. Should symbols for the metro build be available pulbically? I just checked the latest build from today and the debugger is reporting it can't get them.
(In reply to Jim Mathies [:jimm] from comment #9) > (In reply to Aki Sasaki [:aki] from comment #8) > > This landed in production yesterday, sorry for the delay in notification. > > Should symbols for the metro build be available pulbically? I just checked > the latest build from today and the debugger is reporting it can't get them. I landed a fix since we had defined packageTests twice: one to set it to True and the other to False. Once this new change makes it to production we should symbols and package tests being generated.
This made it to production today.
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → FIXED
(In reply to Jim Mathies [:jimm] from comment #9) > Should symbols for the metro build be available pulbically? I just checked > the latest build from today and the debugger is reporting it can't get them. If you mean the symbol server here then that will require enabling nightlies. Dep builds only create the crashreporter-symbols.zip for upload beside the bits, eg at the url Armen gave in comment #12.
(In reply to Nick Thomas [:nthomas] from comment #13) > (In reply to Jim Mathies [:jimm] from comment #9) > > Should symbols for the metro build be available pulbically? I just checked > > the latest build from today and the debugger is reporting it can't get them. > > If you mean the symbol server here then that will require enabling > nightlies. Dep builds only create the crashreporter-symbols.zip for upload > beside the bits, eg at the url Armen gave in comment #12. What would it take to get those going with pdbs served up by the symbol server? The symbol files being zipped up for crashreporter are packaged in a format that is incompatible with Visual Studio.
Ted, do we have space on the symbol server to accommodate uploading dep builds there ? The alternative is something like this: diff --git a/toolkit/mozapps/installer/packager.mk b/toolkit/mozapps/installer/packager.mk --- a/toolkit/mozapps/installer/packager.mk +++ b/toolkit/mozapps/installer/packager.mk @@ -979,17 +979,21 @@ CHECKSUM_FILES = $(CHECKSUM_FILE) UPLOAD_FILES= \ $(call QUOTED_WILDCARD,$(DIST)/$(PACKAGE)) \ $(call QUOTED_WILDCARD,$(INSTALLER_PACKAGE)) \ $(call QUOTED_WILDCARD,$(DIST)/$(COMPLETE_MAR)) \ $(call QUOTED_WILDCARD,$(DIST)/$(LANGPACK)) \ $(call QUOTED_WILDCARD,$(wildcard $(DIST)/$(PARTIAL_MAR))) \ $(call QUOTED_WILDCARD,$(DIST)/$(PKG_PATH)$(TEST_PACKAGE)) \ +ifdef MOZ_METRO + $(call QUOTED_WILDCARD,$(DIST)/$(PKG_PATH)$(SYMBOL_FULL_ARCHIVE_BASENAME).zip) \ +else $(call QUOTED_WILDCARD,$(DIST)/$(PKG_PATH)$(SYMBOL_ARCHIVE_BASENAME).zip) \ +endif $(call QUOTED_WILDCARD,$(DIST)/$(SDK)) \ $(call QUOTED_WILDCARD,$(MOZ_SOURCESTAMP_FILE)) \ $(call QUOTED_WILDCARD,$(PKG_JSSHELL)) \ $(if $(UPLOAD_EXTRA_FILES), $(foreach f, $(UPLOAD_EXTRA_FILES), $(wildcard $(DIST)/$(f)))) SIGN_CHECKSUM_CMD= ifdef MOZ_SIGN_CMD # If we're signing with gpg, we'll have a bunch of extra detached signatures to Which would get us uploading a crashreporter.zip with .pdb's as well as .sym files.
(In reply to Nick Thomas [:nthomas] from comment #15) > Ted, do we have space on the symbol server to accommodate uploading dep > builds there ? The alternative is something like this: I don't think so, no. We've been skating on thin ice there for some time.
Can we do the alternative? I'm just interested in getting at pdbs for builds here - http://ftp.mozilla.org/pub/mozilla.org/firefox/tinderbox-builds/elm-win32-metro/ so that I can debug these builds locally. I don't mind downloading the pdbs along with the zip install to do it.
Assuming that patch actually works (haven't tested it), you can land it in Elm. How long would you want that to last though, and when might Elm merge back to m-c ?
(In reply to Nick Thomas [:nthomas] from comment #18) > Assuming that patch actually works (haven't tested it), you can land it in > Elm. How long would you want that to last though, and when might Elm merge > back to m-c ? merging elm to mc is blocked on bug 755724. Once that's fixed we'll finish up the migration. Maybe within the next couple months. Currently on elm we have both vc10 and experimental vc11 builds going. At some point after the mc merge we'll need elm nightlies for the vc11 builds with symbols uploaded to the symbol server. At which point I guess this could be undone. Does this make sense?
Product: mozilla.org → Release Engineering
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: