Modify devserver and gmerge to be friendlier

Modify devserver and gmerge to be friendlier

1.  If devserver detects that you are trying to build a cros_workon
package that isn't actually cros_workon'd, it will fail.  This should
prevent an infrequent, but highly frustrating error case.

2.  You can supply USE flags on the gmerge command line

3.  Some unit tests added for gmerge

BUG=N0NE
TEST=unit tests + manual testing

Change-Id: I6dacdf16496efb90e18fba7a412339051eebc4e4

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=7c79a81

[rochberg, 2011-01-14 22:31:30]

PTAL

[sosa, 2011-01-14 23:03:54]

+rtc in case he wants to take a look at it.

http://codereview.chromium.org/6305004/diff/2001/devserver.py
File devserver.py (right):

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode84
devserver.py:84: 
Don't need this extra line

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode92
devserver.py:92: 
2 lines between top level defs

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode93
devserver.py:93: def _IsIn(needle, haystack_list):
Isn't this equivalent to needle in haystick_list?

I just tried with 
a="""tacos are super
delicious taco t
blake"""

"taco" in a = True
"blake" in a = True
"alskdjgasljkg" in a = False

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode98
devserver.py:98: 
2 lines between top level defs

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode99
devserver.py:99: class Builder(object):
Should probably move this into its own module.

http://codereview.chromium.org/6305004/diff/2001/gmerge
File gmerge (right):

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode1
gmerge:1: #!/usr/bin/env python
Never seen this style before ...

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode5
gmerge:5: # found in the LICENSE file.
Maybe add a module docstring.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode16
gmerge:16: PARSER = optparse.OptionParser()
Why did you choose to make this global rather than have this in your main?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode16
gmerge:16: PARSER = optparse.OptionParser()
usage?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode45
gmerge:45: def RemountOrChangeRoot(self, environ):
Please add at least one line doc strings to each method.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode57
gmerge:57: ignore_errors=True)
This is somewhat functionally different than the other.  Not sure if it matters
though.  Why did you choose to ignore_errors?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode93
gmerge:93: print result.read()
result.close()?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode98
gmerge:98: sys.exit(1)
sys.exit(e.read())?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode104
gmerge:104: if len(remaining_arguments) != 1:
So this removes the ability to to do gmerge -n -h?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode105
gmerge:105: sys.exit('Usage: gmerge package_name')
It makes it a little easier to read to have a line after conditional blocks.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode110
gmerge:110: merger = GMerger(open('/etc/lsb-release').readlines())
Why not just pass the path?

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode120
gmerge:120: subprocess.call(['emerge', '--getbinpkgonly', '--usepkgonly',
package_name])
check_call?

[rochberg, 2011-01-15 06:16:47]

Please take another look, especially at the binary package cache 'clear' and the
emerge passthrough stuff.

Thx

http://codereview.chromium.org/6305004/diff/2001/devserver.py
File devserver.py (right):

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode84
devserver.py:84: 
On 2011/01/14 23:03:54, sosa wrote:
> Don't need this extra line

Done.

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode93
devserver.py:93: def _IsIn(needle, haystack_list):
The way the code is structured now, the relevant test would be
"taco" in [a], which is False.  But we don't make any use of the line-oriented
nature of the output, so I'll switch to just plain "in" and the raw output

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode98
devserver.py:98: 
On 2011/01/14 23:03:54, sosa wrote:
> 2 lines between top level defs

Done.

http://codereview.chromium.org/6305004/diff/2001/devserver.py#newcode99
devserver.py:99: class Builder(object):
On 2011/01/14 23:03:54, sosa wrote:
> Should probably move this into its own module.

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge
File gmerge (right):

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode1
gmerge:1: #!/usr/bin/env python
On 2011/01/14 23:03:54, sosa wrote:
> Never seen this style before ... 

This is so we can run even if python is in /usr/local/bin

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode5
gmerge:5: # found in the LICENSE file.
On 2011/01/14 23:03:54, sosa wrote:
> Maybe add a module docstring.

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode16
gmerge:16: PARSER = optparse.OptionParser()
On 2011/01/14 23:03:54, sosa wrote:
> Why did you choose to make this global rather than have this in your main?

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode16
gmerge:16: PARSER = optparse.OptionParser()
On 2011/01/14 23:03:54, sosa wrote:
> usage?

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode45
gmerge:45: def RemountOrChangeRoot(self, environ):
On 2011/01/14 23:03:54, sosa wrote:
> Please add at least one line doc strings to each method.

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode57
gmerge:57: ignore_errors=True)
First of all, the ignore_errors bit was due to some confusion and clearly wrong.

But I don't think the line in the original is correct either:

# Delete the local binary package cache.
rm -rf "${PKGDIR}/packages"

There is no such file on my machine.  There are a bunch of directories of tbz2's
and a file Packages (note capitalization) that contains metadata.  

Furthermore, it seems like --getbinpkgonly obviates the need for this, though my
understanding of this is incomplete. 

I'm tentatively removing the whole thing because I don't think it did anything,
but I would like a second opinion.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode93
gmerge:93: print result.read()
On 2011/01/14 23:03:54, sosa wrote:
> result.close()?

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode98
gmerge:98: sys.exit(1)
On 2011/01/14 23:03:54, sosa wrote:
> sys.exit(e.read())?

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode104
gmerge:104: if len(remaining_arguments) != 1:
On 2011/01/14 23:03:54, sosa wrote:
> So this removes the ability to to do gmerge -n -h?

gmerge --help works, but gmerge -n -h as help for emerge does not work.   If
there's a use case where we want to pass nontrivial extra arguments to emerge,
I'm happy to change, but it seems like those seeking help for emerge can run
emerge -h.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode105
gmerge:105: sys.exit('Usage: gmerge package_name')
On 2011/01/14 23:03:54, sosa wrote:
> It makes it a little easier to read to have a line after conditional blocks.

Done.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode110
gmerge:110: merger = GMerger(open('/etc/lsb-release').readlines())
On 2011/01/14 23:03:54, sosa wrote:
> Why not just pass the path?

Easier to test.

http://codereview.chromium.org/6305004/diff/2001/gmerge#newcode120
gmerge:120: subprocess.call(['emerge', '--getbinpkgonly', '--usepkgonly',
package_name])
On 2011/01/14 23:03:54, sosa wrote:
> check_call?

Done.

[sosa, 2011-01-18 19:12:32]

LGTM.

You should remove the RDEPEND on wget from the ebuild and add a RDEPEND on
dev-lang/python