Re: [gnulib-tool-py] Deprecating of rsync and wget

[Bruno Haible]

Hi Dmitriy,

> I'm currently working on generation of po files and recently I've found
> a piece of code where we decide whether to choose between rsync and wget.
> I'd like to deprecate them both since we have some modules in Python which
> allow to download files without calling system utilities (e.g. httplib2).
> How do you look if I deprecate rsync and wget completely? I'd like to
> remove as much bash code as it is possible at this time, but if you want
> I can leave this part of code unchanged.

Regarding rsync, I would prefer to keep it in the code. The
Translation Project has an rsync daemon running, so that we can download
the PO files with compression (and thus 4 x reduced bandwidth); it would
be stupid not to use it.

Regarding wget versus httplib2, it depends how much effort it is to write
the code entirely inside Python. What wget does it that it accesses
http://translationproject.org/latest/gnulib
extracts the hyperlinks to the PO files from this page and downloads them.

Also, it depends whether httplib2 is part of normal Python 3.x installations.
So far, running gnulib-tool uses only standard Python packages, therefore
users of gnulib-tool need to install only Python and no additional Python
packages. Therefore I think http://docs.python.org/py3k/library/http.client.html
and http://docs.python.org/py3k/library/html.parser.html would be better
than httplib2.

Bruno