This started out as just adding __XPC_DYLD_FRAMEWORK_PATH to bisect-builds, but then I tried to share more code in webkitdirs.pm by creating a common "environment setup" subroutine, but the extracted code also set up guardmalloc which seemed useful*, and it seemed silly to manually add '-g|--guard-malloc' switch parsing and help text to every script, so I added methods to make it easy to parse and print help for shared switching (initially just -g|--guard-malloc). In the mid-term, I think we should investigate a WebKit::Options[Parser] class that wraps Getopt::Long, knows about sets of common switches, and knows how to print help by passing it a list of switches for a particular script. * Already found a bug in WebKit2.CookieManager when run under guardmalloc. (Not a crash, but test expectation fails.)
Created attachment 231517 [details] Patch v1
(In reply to comment #0) > * Already found a bug in WebKit2.CookieManager when run under guardmalloc. (Not a crash, but test expectation fails.) I filed this as Bug 132955.
Comment on attachment 231517 [details] Patch v1 Clearing flags on attachment: 231517 Committed r169042: <http://trac.webkit.org/changeset/169042>
All reviewed patches have been landed. Closing bug.
Did this break run-api-tests? http://build.webkit.org/builders/Apple%20MountainLion%20Debug%20WK1%20%28Tests%29/builds/14033/steps/run-api-tests/logs/stdio
(In reply to comment #5) > Did this break run-api-tests? > http://build.webkit.org/builders/Apple%20MountainLion%20Debug%20WK1%20%28Tests%29/builds/14033/steps/run-api-tests/logs/stdio Speculative fix: Committed r169078: <http://trac.webkit.org/changeset/169078>