Comment 12 for bug 541990

On Mar 31, 2010, at 03:51 PM, Martin Pitt wrote:

>> Yes, the front-end currently throws DBusExceptions. What should it do
>instead?
>
>I guess it should handle them gracefully, like restarting the backend or
>crashing the UI as well, since otherwise the user will be left with a
>nonfunctioning UI. With such a lot of code changes, some exceptions will
>be inevitable, I figure.

I think dbus should probably be responsible for restarting the backend, but I
can see where the frontends should catch dbus exceptions and exit more
cleanly.

>> * Landmark package {0.package} is missing
>> * Landmark package {0.package} is not downloadable
>> * Duplicate cruft with different cleanup: {0.cruft_name}
>> * No such cruft: {0.cruft_name}
>> - Error messages that actually shouldn't be translated!
>
>Right, are they marked as such? Does the user actually see them? User-
>visible text should not use geek terminology like "landmark" (what is
>that even, in a packaging context?) or "cruft".

They're marked for translation in trunk, but I've removed the markings in
lp:~barry/computer-janitor/ppa -- that branch is currently up for review, but
I am going to modify the branch to include resolution of the above issue, as
well as...

>> Modified:
>
>CLI help is fine (although it should boil down to a few removed and
>added strings if you use optparse properly).

Actually, I'm using argparse 1.1 (which has been given a MIR and an FFe and is
just waiting for cj 2.0 to land), but I will take a look at whether the help
can be adjusted to be more like the previous help.

>Can we please avoid the other string changes? Each of these breaks all
>the translations which have been done by now. These might be moved to
>lucid+1?

I think the other three, while improvements, are not enough to push for in
Lucid, so I'll revert them and create a branch to add them back for lucid+1.