duplicity-team team mailing list archive

[Kenneth Loafman <kenneth@xxxxxxxxxxx>]

Yes, had some fun with that merge and bzr this time around, but that was a
background problem.

There was one of MS's smart (dumb) chars as a dash in one of the comments,
a '--' turned into '-', but slightly wider.  Hard to spot, and a problem
for Python, even in a comment.

The thing that surprised me is that neither pylint or pycodestyle caught
it.  Ah well.

...Ken


On Fri, May 5, 2017 at 9:32 AM, Aaron <lists@xxxxxxxxxxxxxxxxxx> wrote:

> Ken,
>
> I see revision 1212 has the explanation "Replace incoming non-ASCII chars
> in commandline.py" but the changes all seem to be about Azure (and look to
> replicate changes already made in 1210?) -- I wonder if you made a mistake?
>
> I am particularly interested in encoding questions because I am currently
> doing a large amount of (unmerged) work on improving unicode support in
> duplicity. Reading the history, it looks as though Michael Terry's Python 3
> branch hit a roadblock with all of the unicode/bytes work, so thought I
> could tackle chunks of that first to ease the transition to Python 3. The
> key advantage there being that I should be able to merge in that work more
> gradually and avoid the work of keeping a parallel Python 3 branch up to
> date for a long conversion period. I am incorporating relevant work from
> Michael's branch as I go. It is thankfully easier now than it would have
> been when Michael started his branch (many libraries happily accept/return
> unicode).
>
> I'll put up a branch shortly. So far I have essentially finished the
> select/glob matching side, but it is not quite ready for merging yet. It is
> working, but still assumes a UTF-8 system in a couple of places and the
> version of pexpect in the old LTS Ubuntus is so old that it has unicode
> bugs to work around.
>
> Kind regards,
>
> Aaron
>
>