Summary: | Add necessary build steps to JavaScriptGlue GYP project. | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Dimitri Glazkov (Google) <dglazkov> | ||||||
Component: | New Bugs | Assignee: | Dimitri Glazkov (Google) <dglazkov> | ||||||
Status: | RESOLVED FIXED | ||||||||
Severity: | Normal | CC: | abarth | ||||||
Priority: | P2 | ||||||||
Version: | 528+ (Nightly build) | ||||||||
Hardware: | Other | ||||||||
OS: | OS X 10.5 | ||||||||
Bug Depends on: | |||||||||
Bug Blocks: | 55018 | ||||||||
Attachments: |
|
Description
Dimitri Glazkov (Google)
2011-02-24 16:54:55 PST
Created attachment 83750 [details]
WIP: Not building yet, capturing progress so far.
Comment on attachment 83750 [details] WIP: Not building yet, capturing progress so far. View in context: https://bugs.webkit.org/attachment.cgi?id=83750&action=review > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:20 > + '../Info.plist', Should this be in JavaScriptGlue.gypi? > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:36 > + 'action': [ 'if [ -f ../../Tools/Scripts/check-for-global-initializers ]; then\n ../../Tools/Scripts/check-for-global-initializers || exit $?\nfi";' ], I'm not sure whether it's better to put these inline in the gyp or to having them be external sh files. In JavaScriptCore.gyp, I went with the external sh files. I'm not sure this matters at this stage. Just thinking out loud. > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:217 > + 'inputs': [ '$(SRCROOT)/Configurations/Version.xcconfig' ], This won't work properly because the current plan is to suck all the xcconfig files into the gyp file. If there's a way to grab the xcconfig files from the file system, that might be better. > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:218 > + 'outputs': [ '$(SRCROOT)/Info.plist' ], Probably better to not use $ variables in inputs and outputs. $ variables are only understood by Xcode. If we use < variables, they'll be understood by gyp. Created attachment 83830 [details]
Patch
(In reply to comment #2) > (From update of attachment 83750 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=83750&action=review > > > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:20 > > + '../Info.plist', > > Should this be in JavaScriptGlue.gypi? Done. > > > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:36 > > + 'action': [ 'if [ -f ../../Tools/Scripts/check-for-global-initializers ]; then\n ../../Tools/Scripts/check-for-global-initializers || exit $?\nfi";' ], > > I'm not sure whether it's better to put these inline in the gyp or to having them be external sh files. In JavaScriptCore.gyp, I went with the external sh files. I'm not sure this matters at this stage. Just thinking out loud. Pulled out. > > > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:217 > > + 'inputs': [ '$(SRCROOT)/Configurations/Version.xcconfig' ], > > This won't work properly because the current plan is to suck all the xcconfig files into the gyp file. If there's a way to grab the xcconfig files from the file system, that might be better. Turns out, it's not necessary for the script. I was just mimicking the original project file. > > > Source/JavaScriptGlue/gyp/JavaScriptGlue.gyp:218 > > + 'outputs': [ '$(SRCROOT)/Info.plist' ], > > Probably better to not use $ variables in inputs and outputs. $ variables are only understood by Xcode. If we use < variables, they'll be understood by gyp. Done. Comment on attachment 83830 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=83830&action=review This is great! Thanks. > Source/JavaScriptGlue/ChangeLog:12 > + * gyp/remove-headers-if-needed.sh: Added. > + * gyp/run-if-exists.sh: Added. > + * gyp/update-info-plist.sh: Added. We'll probably want to move these up in the source tree so they can be shared by multiple projects, but we can do that once we tackle the sharing problems. > Source/JavaScriptGlue/gyp/run-if-exists.sh:4 > +if [ -f ../../Tools/Scripts/$1 ]; then > + ../../Tools/Scripts/$1 || exit $?; It would be nice if we didn't have to hard-code the "../.." here. Maybe run-if-exists should take the fully relative path and then we can let GYP magic figure out the depth? Committed r79890: <http://trac.webkit.org/changeset/79890> |