win: Test some basic ! windbg commands

win: Test some basic ! windbg commands

R=mark@chromium.org
BUG=crashpad:20, crashpad:46, crashpad:52
Committed: https://chromium.googlesource.com/crashpad/crashpad/+/bbd00c3a91fac9a25a0351337602e6477987a7b7

[scottmg, 2015-10-09 17:39:31]

Updated to just use subprocess and re directly. pexpect was overkill.

I've also installed cdb on the x64 bot, so this should work on the waterfall
now.

[Mark Mentovai, 2015-10-09 18:34:05]

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
File snapshot/win/end_to_end_test.py (right):

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:40: search_paths =
[os.getenv('PROGRAMFILES(X86)'), os.getenv('PROGRAMFILES'),
I liked this better before, one element per line. Same for possible_paths below.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:80: pipe_name = r'\\.\pipe\end-to-end'
Should this be a little more random? pid + 64 random bits ought to be enough.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:93: pipe_name, '--database=' + test_database
'pipe_name=' and pipe_name should at least be on the same line. I think it’s
preferable to break at commas than at pluses.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:115: `Check()` searchs for regex patterns in
sequence allowing verification of
searches

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:122: ';'.join(commands) + ';q'])
'-c' next to this, since this whole thing applies to the -c.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:122: ';'.join(commands) + ';q'])
Dunno about how cdb command lines should be escaped. I guess this is OK since
it’s so purpose-specific and the set of commands are within our control. In
fact, you never even invoke this with more than one command, and you’ve got a
comment about not doing that below, so is the join just a vestige from a
previous version of this?

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:131: print '\rok - '
'\rok - '

Not really sure what you’re going for here.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:135: print '-'*80
Errors to stderr?

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:184: print 'must supply out dir'
stderr for these

[scottmg, 2015-10-09 18:49:28]

Thanks

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
File snapshot/win/end_to_end_test.py (right):

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:40: search_paths =
[os.getenv('PROGRAMFILES(X86)'), os.getenv('PROGRAMFILES'),
On 2015/10/09 18:34:04, Mark Mentovai wrote:
> I liked this better before, one element per line. Same for possible_paths
below.

Yeah, I started using https://github.com/google/yapf in --style=chromium but
it's not as reliable as clang-format. I'll have to go back to manual formatting
I think.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:80: pipe_name = r'\\.\pipe\end-to-end'
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> Should this be a little more random? pid + 64 random bits ought to be enough.

Done.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:93: pipe_name, '--database=' + test_database
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> 'pipe_name=' and pipe_name should at least be on the same line. I think it’s
> preferable to break at commas than at pluses.

Done.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:115: `Check()` searchs for regex patterns in
sequence allowing verification of
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> searches

Done.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:122: ';'.join(commands) + ';q'])
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> '-c' next to this, since this whole thing applies to the -c.

Done.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:122: ';'.join(commands) + ';q'])
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> Dunno about how cdb command lines should be escaped. I guess this is OK since
> it’s so purpose-specific and the set of commands are within our control. In
> fact, you never even invoke this with more than one command, and you’ve got a
> comment about not doing that below, so is the join just a vestige from a
> previous version of this?

Switched to just 'command'.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:131: print '\rok - '
On 2015/10/09 18:34:04, Mark Mentovai wrote:
> '\rok - '
> 
> Not really sure what you’re going for here.

I wanted the "ok"s before the names so they're easier to read for normal output,
but I also want to have the message printed at failure time.

But... it only made sense when actual work was happening during the expect.
Since it's just an re search now it can be simpler.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:135: print '-'*80
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> Errors to stderr?

Done.

https://codereview.chromium.org/1397833004/diff/80001/snapshot/win/end_to_end...
snapshot/win/end_to_end_test.py:184: print 'must supply out dir'
On 2015/10/09 18:34:05, Mark Mentovai wrote:
> stderr for these

Done.

[Mark Mentovai, 2015-10-09 18:57:13]

LGTM

[scottmg, 2015-10-09 20:39:44]

Committed patchset #6 (id:100001) manually as
bbd00c3a91fac9a25a0351337602e6477987a7b7 (presubmit successful).