profile: make the contextmanager object available to the callers
This will allow calling methods on the object in the code using the context
manager.
profile: introduce a knob to control if the context is actually profiling
This is a step toward allowing context where the profiling in enabled
withing the context range.
This also open the way to kill the dedicated "maybeprofile" context manager
and keep only one of 'profile' and 'maybeprofile'.
profile: introduce a "start" method to the profile context
The start method is doing all profiler setup and activation. It is currently
unconditionally called by '__init__' but this will be made more flexible in
later changesets.
profile: upgrade the "profile" context manager to a full class
So far we have been able to use a simple decorator for this. However using the
current context manager makes the scope of the profiling in dispatch
constrainted and the time frame to decide to enable profiling quite limited
(using "maybeprofile")
This is the first step toward the ability to enable the profiling from within
the profiling scope. eg::
with maybeprofiling(ui) as profiler:
...
bar.foo():
...
if options['profile']:
profiler.start()
...
fooz()
...
My target usecase is adding support for "--profile" to alias definitions with
effect. These are to be used with "profiling.output=blackbox" to gather data
about operation that get slow from time to time (eg: pull being minutes instead
of seconds from time to time).
Of course, in such case, the scope of the profiling would be smaller since
profiler would be started after running extensions 'reposetup' (and other
potentially costly logic), but these are not relevant for my target usecase
(multiple second commits, multiple tens of seconds pull).
Currently adding '--profile' to a command through alias requires to re-spin a
Mercurial binary (using "!$HG" in alias), which as a significant performance
impact, especially in context where startup performance is being worked on...
An alternative approach would be to stop using the context manager in dispatch
and move back to a try/finally setup.
setup: avoid linker warnings on Windows about multiple export specifications
The PyMODINIT_FUNC macro contains __declspec(dllexport), and then the build
process adds an "/EXPORT func" to the command line. The 64-bit linker flags
this [1].
Everything except zstd.c and bser.c are covered by redefining the macro in
util.h [2]. These modules aren't built with util.h in the #include path, so the
redefining hack would have to be open coded two more times.
After seeing that extra_linker_flags didn't work, I couldn't find anything
authoritative indicating why, though I did see an offhand comment on SO that
CFLAGS is also ignored on Windows. I also don't fully understand the
interaction between msvccompiler and msvc9compiler- I first subclassed the
latter, but it isn't used when building with VS2008.
I know the camelcase naming isn't the standard, but the HackedMingw32CCompiler
class above it was introduced 5 years ago (and I think the current style was
in place by then), so I assume that there's some reason for it.
[1] https://support.microsoft.com/en-us/help/835326/you-receive-an-lnk4197-error-in-the-64-bit-version-of-the-visual-c-compiler
[2] https://bugs.python.org/
issue9709#msg120859
memctx: always use cache for filectxfn
I don't see a downside to doing this unless I'm missing something.
Thanks to foozy for correcting my previous bad logic.
test-hardlinks: stabilize for Windows
This broke in
c2cb0de25120, which breaks hardlinks when the executable bit is
toggled.