Open
Bug 683397
Opened 14 years ago
Updated 3 years ago
leaktest.py.in should not be interpolated
Categories
(Firefox Build System :: General, defect)
Firefox Build System
General
Tracking
(Not tracked)
NEW
People
(Reporter: k0scist, Unassigned)
References
Details
(Whiteboard: [mozbase])
Attachments
(1 file, 2 obsolete files)
1.88 KB,
patch
|
ted
:
review+
|
Details | Diff | Splinter Review |
http://mxr.mozilla.org/mozilla-central/source/build/leaktest.py.in
The only thing interpolated seems to be the top line,
#literal #!/usr/bin/python
which if it wasn't an interpolated .in file wouldn't be necessary (though technically #!/usr/bin/env python is a more proper she-bang)
Reporter | ||
Comment 1•14 years ago
|
||
I'm not seeing where leaktest is run at all, but this seems to fix all references to it I could find. Has not yet been run on try
Assignee: nobody → jhammel
Attachment #559915 -
Flags: review?(ted.mielczarek)
Comment 2•14 years ago
|
||
It gets run directly by the buildbot code somewhere. I think as long as you copy it to the same place you should be ok.
Comment 3•14 years ago
|
||
We could probably update Buildbot to look for it in the new location, too. But having it copied to the old is basically necessary as a transitional piece to avoid burning.
Comment 4•14 years ago
|
||
As is remembering to clobber, everything on every branch as it merges around, plus telling fifty people "you need to clobber" on IRC.
Is it worth it?
Comment 5•14 years ago
|
||
Why is a clobber needed?
Reporter | ||
Comment 6•14 years ago
|
||
My intention (which I think I achieved) is that leaktest.py would be in the same location (and copied over as well). Did I somehow mess this up? Given that, I don't really understand comments 3 and 4.
Comment 7•14 years ago
|
||
Oh right. Every time we fix one of these we get bitten by the fact that Python's sys.path entry for "the directory containing the script being run" does not do what you expect with symlinks. It's the directory of the *target*. nsinstall symlinks by default on POSIX, so this breaks because you can't import automation.py. (It wasn't broken before because you preprocessed the file so it wasn't a symlink.)
You can fix this by setting NSDISTMODE = copy, but it's possible that that also screws something up, as philor is saying. The details are fuzzy.
Comment 8•14 years ago
|
||
Comment on attachment 559915 [details] [diff] [review]
patch
Review of attachment 559915 [details] [diff] [review]:
-----------------------------------------------------------------
You definitely need an NSDISTMODE = copy in the Makefile somewhere. I don't know that this will work properly after that, we'd have to go look back at the bugs where we did this previously and see what we wound up doing.
Attachment #559915 -
Flags: review?(ted.mielczarek) → review-
Updated•14 years ago
|
Product: Firefox → Core
QA Contact: build.config → build-config
Reporter | ||
Comment 9•14 years ago
|
||
Attachment #559915 -
Attachment is obsolete: true
Attachment #569540 -
Flags: review?(ted.mielczarek)
Comment 10•14 years ago
|
||
Comment on attachment 569540 [details] [diff] [review]
update with NSDISTMODE = copy
Review of attachment 569540 [details] [diff] [review]:
-----------------------------------------------------------------
::: build/Makefile.in
@@ +93,5 @@
> $(topsrcdir)/build/pgo/blueprint/grid.png \
> $(topsrcdir)/build/pgo/blueprint/fancytype-screen.css \
> $(NULL)
>
> +libs:: NSDISTMODE = copy
No, you don't want to do this. Just put NSDISTMODE = copy as an assignment at the top-level of the Makefile, not as a target-specific variable.
Attachment #569540 -
Flags: review?(ted.mielczarek) → review-
Reporter | ||
Comment 11•14 years ago
|
||
Why shouldn't it be a target-specific variable? Its an easy change, i just didn't want to affect the rest of the file.
Reporter | ||
Comment 12•14 years ago
|
||
Attachment #569540 -
Attachment is obsolete: true
Attachment #569820 -
Flags: review?(ted.mielczarek)
Updated•14 years ago
|
Attachment #569820 -
Flags: review?(ted.mielczarek) → review+
Reporter | ||
Comment 13•14 years ago
|
||
I don't have push access to m-c. Ted, could you please push this for me?
Reporter | ||
Updated•14 years ago
|
Whiteboard: [mozbase] → [mozbase][checkin-needed]
Reporter | ||
Updated•14 years ago
|
Keywords: checkin-needed
Whiteboard: [mozbase][checkin-needed] → [mozbase]
Comment 14•14 years ago
|
||
Has this been sent to try? If not, I'll include it next time I send stuff there.
Also, please may you include a commit message. Thanks :-)
Status: NEW → ASSIGNED
Reporter | ||
Comment 15•14 years ago
|
||
commit message should probably be: "Bug 683397 - leaktest.py.in should not be interpolated" Will test on try now
Reporter | ||
Comment 16•14 years ago
|
||
On try, all builds fail with messages like http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/jhammel@mozilla.com-eb7bc8d562ff/try-macosx-debug/try-macosx-debug-build4305.txt.gz:
make -C unix libs
/tools/python/bin/python2.5 /builds/slave/try-osx-dbg/build/config/Preprocessor.py \
-DBROWSER_PATH=\"../dist/NightlyDebug.app/Contents/MacOS/firefox-bin\" -DXPC_BIN_PATH=\"/builds/slave/try-osx-dbg/build/obj-firefox/dist/bin\" -DBIN_SUFFIX=\"\" -DPROFILE_DIR=\"../_profile/pgo\" -DCERTS_SRC_DIR=\"/builds/slave/try-osx-dbg/build/build/pgo/certs\" -DPERL="\"/usr/bin/perl\"" -DIS_MAC=1 -DIS_LINUX=0 -DIS_CAMINO=0 -DIS_TEST_BUILD=1 -DIS_DEBUG_BUILD=1 -DCRASHREPORTER=1 -DOSTYPE=\"Darwin9.2.0\" -DOSARCH=Darwin -DMOZILLA_VERSION=\"10.0a1\" -DMOZILLA_VERSION_U=10.0a1 -DXP_MACOSX=1 -DXP_DARWIN=1 -DD_INO=d_ino -DSTDC_HEADERS=1 -DHAVE_SSIZE_T=1 -DHAVE_ST_BLKSIZE=1 -DHAVE_SIGINFO_T=1 -DHAVE_INT16_T=1 -DHAVE_INT32_T=1 -DHAVE_INT64_T=1 -DHAVE_UINT=1 -DHAVE_VISIBILITY_HIDDEN_ATTRIBUTE=1 -DHAVE_VISIBILITY_ATTRIBUTE=1 -DHAVE_DIRENT_H=1 -DHAVE_GETOPT_H=1 -DHAVE_MEMORY_H=1 -DHAVE_UNISTD_H=1 -DHAVE_NL_TYPES_H=1 -DHAVE_X11_XKBLIB_H=1 -DHAVE_SYS_STATVFS_H=1 -DHAVE_MMINTRIN_H=1 -DHAVE_SYS_CDEFS_H=1 -DHAVE_DLADDR=1 -DFUNCPROTO=15 -DHAVE_LIBXSS=1 -DHAVE_FT_BITMAP_SIZE_Y_PPEM=1 -DHAVE_FT_GLYPHSLOT_EMBOLDEN=1 -DHAVE_FT_LOAD_SFNT_TABLE=1 -DHAVE_FT_SELECT_SIZE=1 -DHAVE_RANDOM=1 -DHAVE_STRERROR=1 -DHAVE_LCHOWN=1 -DHAVE_FCHMOD=1 -DHAVE_SNPRINTF=1 -DHAVE_MEMMOVE=1 -DHAVE_RINT=1 -DHAVE_STAT64=1 -DHAVE_LSTAT64=1 -DHAVE_SETBUF=1 -DHAVE_ISATTY=1 -DHAVE_FLOCKFILE=1 -DHAVE_LOCALTIME_R=1 -DHAVE_STRTOK_R=1 -DHAVE_LANGINFO_CODESET=1 -DVA_COPY=va_copy -DHAVE_VA_COPY=1 -DMALLOC_H=\<malloc/malloc.h\> -DHAVE_I18N_LC_MESSAGES=1 -DHAVE_LOCALECONV=1 -DNS_ALWAYS_INLINE=__attribute__\(\(always_inline\)\) -DNS_ATTR_MALLOC=__attribute__\(\(malloc\)\) -DNS_WARN_UNUSED_RESULT=__attribute__\(\(warn_unused_result\)\) -DNS_NORETURN=__attribute__\(\(noreturn\)\) -DMOZ_PHOENIX=1 -DMOZ_BUILD_APP=browser -DMOZ_WIDGET_COCOA=1 -DMOZ_INSTRUMENT_EVENT_LOOP=1 -DMOZ_DISTRIBUTION_ID=\"org.mozilla\" -DIBMBIDI=1 -DNS_PRINTING=1 -DNS_PRINT_PREVIEW=1 -DMOZ_OGG=1 -DATTRIBUTE_ALIGNED_MAX=64 -DMOZ_WEBM=1 -DVPX_X86_ASM=1 -DMOZ_WAVE=1 -DMOZ_SYDNEYAUDIO=1 -DMOZ_MEDIA=1 -DMOZ_VORBIS=1 -DMOZ_XTF=1 -DENABLE_SYSTEM_EXTENSION_DIRS=1 -DMOZ_CRASHREPORTER=1 -DMOZ_CRASHREPORTER_ENABLE_PERCENT=100 -DLIBJPEG_TURBO_X86_ASM=1 -DMOZ_UPDATER=1 -DMOZ_UPDATE_CHANNEL=default -DMOZ_FEEDS=1 -DMOZ_SAFE_BROWSING=1 -DMOZ_URL_CLASSIFIER=1 -DMOZ_DEBUG_SYMBOLS=1 -DMOZ_LOGGING=1 -DNS_TRACE_MALLOC=1 -DHAVE___CXA_DEMANGLE=1 -DMOZ_DEMANGLE_SYMBOLS=1 -DHAVE__UNWIND_BACKTRACE=1 -DJS_DEFAULT_JITREPORT_GRANULARITY=3 -DMOZ_OMNIJAR=1 -DMOZ_USER_DIR=\"Mozilla\" -DMOZ_STATIC_JS=1 -DHAVE_STDINT_H=1 -DHAVE_INTTYPES_H=1 -DMOZ_TREE_CAIRO=1 -DHAVE_UINT64_T=1 -DMOZ_TREE_PIXMAN=1 -DMOZ_XUL=1 -DMOZ_PROFILELOCKING=1 -DMOZ_RDF=1 -DBUILD_CTYPES=1 -DMOZ_PLACES=1 -DMOZ_SERVICES_SYNC=1 -DMOZ_APP_UA_NAME=\"\" -DMOZ_APP_UA_VERSION=\"10.0a1\" -DMOZ_UA_FIREFOX_VERSION=\"10.0a1\" -DFIREFOX_VERSION=10.0a1 -DMOZ_UA_BUILDID=\"\" -DMOZ_DLL_SUFFIX=\".dylib\" -DXP_UNIX=1 -DUNIX_ASYNC_DNS=1 -DMOZ_REFLOW_PERF=1 -DMOZ_REFLOW_PERF_DSP=1 /builds/slave/try-osx-dbg/build/build/automation.py.in > automation.py
make[5]: *** No rule to make target `leaktest.py'. Stop.
make[5]: *** Waiting for unfinished jobs....
I do not get this for a local build and don't really understand why I have any target that looks for leaktest.py after the changes I've made above.
This seems to be some subtlty in the build system, as least as setup on try. Anyone have a clue what's going on here? (And for bonus points, how to fix it?)
Updated•14 years ago
|
Keywords: checkin-needed
Reporter | ||
Comment 17•13 years ago
|
||
Not working on this at the moment so unassigning.
Assignee: jhammel → nobody
Updated•7 years ago
|
Product: Core → Firefox Build System
Comment 19•7 years ago
|
||
No assignee, updating the status.
Updated•3 years ago
|
Severity: normal → S3
You need to log in
before you can comment on or make changes to this bug.
Description
•