Fixed bug related to dirty workdir

[dbrgn]

• describe_out[-1] may contain a newline, therefore the strip() is necessary.
• is does identity comparison. You need == there.

[peritus]

Thank you for the fix. Just out of curiosity, in which situation is this necessary ?

[dbrgn]

In the situation where the entire program crashes because my describe_out[-1] looks like dirty\n ;)

Because then it's not stripped away and the distance_to_latest_tag assignment tries to convert a commit hash to an integer.

[peritus]

I want to understand how describe_out[-1] could ever look like dirty\n — is it a specific git configuration or an old version or something ? Do you have a stacktrace or a way to reproduce it ?

Also, I'd like to build a regression test against this before merging.

[dbrgn]

Well, you're running a bash command in the background. Usually any command output appends a newline at the end. Might be a git version issue.

$ git --version
git version 1.8.5.1
$ uname -a
Linux t410 3.12.3-1-ARCH #1 SMP PREEMPT Wed Dec 4 21:45:42 CET 2013 x86_64 GNU/Linux

And describe_out:

$ python
Python 2.7.6 (default, Nov 26 2013, 12:52:49) 
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import subprocess
>>> describe_out = subprocess.check_output([
... "git",
... "describe",
... "--dirty",
... "--tags",
... "--long",
... "--abbrev=40",
... "--match=v*",
... ], stderr=subprocess.STDOUT)
>>> describe_out
'v0.3.6-1-gd9527ff2b3d409abec266ac99dbacc5faed2e1b2\n'

[peritus]

Thanks, that explains it.

That line of code is never reached because at the moment at dirty workdir aborts bumpversion very early on. I eventually want to loosen that restriction, so: thanks for finding this before anyone could execute these lines.

[dbrgn]

Actually it was reached (when my directory was dirty), otherwise it wouldn't have crashed bumpversion. But I never investigated how it reached that line.

But in any case, this is now fixed and shouldn't be a problem anymore.

[peritus]

Right. Thanks again.

Sent from my pocket calculator

On 13 Dec 2013, at 15:45, Danilo Bargen [email protected] wrote:

Actually it was reached (when my directory was dirty), otherwise it wouldn't have crashed bumpversion. But I never investigated how it reached that line.

But in any case, this is now fixed and shouldn't be a problem anymore.

—
Reply to this email directly or view it on GitHub.

[peritus]

This is now part of v0.3.8. Running pip install --upgrade bumpversion should get you there.

Also: I'm sorry for realising too late that we could have met at 30C3 to nerd out about bumping versions :(

[dbrgn]

Great :)

Also: I'm sorry for realising too late that we could have met at 30C3 to nerd out about bumping versions :(

Ah. Next time maybe :)