Now that defining $DMD is no longer necessary to run DMD, this patch does the
following.
- Removes all the places where we set DMD=1 (test harnesses, etc.)
- Still handles DMD=1, for backwards compatibility.
- Prints "$DMD is undefined" at DMD start-up if appropriate.
- Writes a |null| value for |dmdEnvVar| in the JSON if $DMD is undefined. Bumps
the DMD output version number accordingly.
- Changes a bunch of the test files accordingly, including changing the mode of
script-ignore-alloc-fns.json in order to test a case where $DMD is undefined.
Various os.path attributes are being used in tight loops. Having local
variables prevents extra dictionary lookups.
This appears to shave 10-20ms off of the tests install manifest
processing time.
FileCopier.copy() was performing a lot of os.path.normpath() operations.
Profiling revealed that os.path.normpath() was the function with the
most wall time CPU usage when processing the tests manifests. Upon
subsequent examination of the code in question, all the paths being used
were already normalized. So, os.path.normpath() wasn't accomplishing
anything.
This patch results in ~300ms reduction in wall time to process the tests
install manifest on a fully populated page cache. Execution time drops
from ~2.8s to ~2.5s.
Profiling reveals that after this patch os.stat() is the #1 wall time
consumer. However, os.path.{join,dirname,normpath} still account for
~1.5x the wall time of os.stat(). There is still room to optimize
this function.
This patch moves profiling mode selection from post-processing (in dmd.py) to
DMD start-up. This will make it easier to add new kinds of profiling, such as
cumulative heap profiling.
Specifically:
- There's a new --mode option. |LiveWithReports| is the default, as it is
currently.
- dmd.py's --ignore-reports option is gone.
- There's a new |mode| field in the JSON output.
- Reports-related operations are now no-ops if DMD isn't in LiveWithReports
mode.
- Diffs are only allowed for output files that have the same mode.
- A new function ResetEverything() replaces the SetSampleBelowSize() and
ClearBlocks(), which were used by the test to change DMD options.
- The tests in SmokeDMD.cpp are split up so they can be run multiple times, in
different modes. The exact combinations of tests and modes has been changed a
bit.
People often seek to learn how mach commands work. A common way to do
this is to launch a debugger and step through the code as it is
executing. But this requires someone to first find and modify the mach
command. This involves overhead.
This patch adds a global --debug-command argument to mach. When present,
we launch an interactive debugger right before command dispatch. This
allows people to easily enter a debugger to see what mach commands are
doing, hopefully lowering the barrier to understanding and contributing.
buildlist invocations are slow and can occur in parallel since the
underlying program obtains a lock on the modified file.
Moving the XPT-related buildlist invocation from the serial libs tier to
the parallel misc tier decreased my no-op build time on OS X from 43.5s
to 37.0s.
When the misc tier was added, only directories with misc-associated
variables from moz.build were traversed. This patch adds a dummy
variable to moz.build whose presence will add the directory to the misc
tier.
This will enable us to aggressively convert existing libs:: rules
to the misc tier.
JS module installation performs simple file copying or preprocessing.
There is no reason it can't occur in parallel. Move it to the misc tier.
As part of this, I recognized that TESTING_JS_MODULES was assigned to a
tier. Since these files are managed by an install manifest, they don't
belong to any tier. So the tier is now listed as None.
The build system being what it currently is, there are various cases where one
wants something explicit, rather than the current autodetection.
For instance, one may want to run
make -C $objdir chrome
instead of
make -C $objdir/chrome
that mach build chrome currently invokes.
There are several such usecases that mach's autodetection makes harder, and
it's sometimes awkward when telling people, to debug their issues, to run
make -C objdir something
and hear back that objdir doesn't exist or something along those lines,
because they took "objdir" literally.
There are, sadly, many combinations of linkage in use throughout the tree.
The main differentiator, though, is between program/libraries related to
Gecko or not. Kind of. Some need mozglue, some don't. Some need dependent
linkage, some standalone.
Anyways, these new templates remove the need to manually define the
right dependencies against xpcomglue, nspr, mozalloc and mozglue
in most cases.
Places that build programs and were resetting MOZ_GLUE_PROGRAM_LDFLAGS
or that build libraries and were resetting MOZ_GLUE_LDFLAGS can now
just not use those Gecko-specific templates.
