Fixed upstream in master and 0.2.x -- I've also added this as a fix to the F9 package; I'll build it this afternoon when I've added a couple more fixes to the build. Thanks for reporting dude.

I'll build when koji isn't down.... :-(

I'm going to reuse this, as it's now a slightly different bug but basically the same problem. It now shows...

Darcs is intended to be an ``advanced'' revision control system

Versiuon is: 

gnome-packagekit.x86_64                  0.3.5-1.fc10                  installed

"Darcs is intended to be an ``advanced'' revision control system." gets pushed through g_markup_escape_text, hence the oddity.

I've made this a bit more sane, and also made the quotes into proper UTF8 quotes, so now the output is "Darcs is intended to be an “advanced” revision control system"

I've fixed this in 57842660963281632327a20f744f22aca9b63cf5, many thanks.

 I've tried to look for that commit on the gnome-packagekit and PackageKit git repos, but I can't find it?

Ahh, it's not synced to the anon repo yet. Do you want access to the developer repo? If so just send me a private mail with your RSA public key and your chosen username and I'll give you access.

Ok, I'm going to reopen this, because the workaround is: 

+       egg_strreplace_indirect (&text, "``", "“");
+       egg_strreplace_indirect (&text, "''", "”");
+       egg_strreplace_indirect (&text, "'", "’");

...this is bad IMO as IMNSHO PK shouldn't be altering the data unless it has to, if we want to change the above package descriptions we should just do that in Fedora.
 A good example of the problem here is python-peak-util-addons in Fedora, as it just uses ``. Likewise I'm sure there are some cases of packages using ' where they really don't want ’.

 So is it possible to fix this without changing the data?

*** Bug 467134 has been marked as a duplicate of this bug. ***

(In reply to comment #8)
> ...this is bad IMO as IMNSHO PK shouldn't be altering the data unless it has
> to, if we want to change the above package descriptions we should just do that
> in Fedora.

Is Fedora policy to allow UTF8 in the package description?

>  A good example of the problem here is python-peak-util-addons in Fedora, as it
> just uses ``. Likewise I'm sure there are some cases of packages using ' where
> they really don't want ’.

The only way I think this would work would be to write a mail to fedora-devel and ask maintainers to clean up the use of ”, ” and ’.

>  So is it possible to fix this without changing the data?

If the packages can be easily fixed, then sure. Otherwise I think it has to continue doing this replacement. Note, it also does replacement this like this when the changelog contains "* " and a bullet is used instead. I can't see packagers using the bullet in changelogs anytime soon.

Richard.

Well, one could think about asking maintainers to stop using those special chars. That would require a lot of work, but it can be done.

However, think about the translations. In a lot of languages, chars like "é" or "ä" are used, and, at least in French, there is a huge difference in meaning between "trompe" and "trompé".

I don't think it is a viable solution to simply prevent the use of those chars.

(In reply to comment #11)
> I don't think it is a viable solution to simply prevent the use of those chars.

Ohh, I _want_ the package maintainers to use UTF8. It's the only sane way of doing things. The problem is a lot of people live in an ASCII world.

> Is Fedora policy to allow UTF8 in the package description?

 Yes, I believe this was mainly so that © and ® could be used ... but as   Mathieu says, this is also required handling for non-en data.

> The only way I think this would work would be to write a mail to fedora-devel
> and ask maintainers to clean up the use of ”, ” and ’.

 Seems reasonable to me.

> >  So is it possible to fix this without changing the data?

> If the packages can be easily fixed, then sure. Otherwise I think it has to
> continue doing this replacement.

What I meant was "is it possible to fix PK to show the data that is there now, without munging it"? While we can probably use rpmlint etc. to eradicate '"' and "'" bytes from Fedora pacakges, they are still going to have to be handled in random packages.


> Note, it also does replacement this like this
> when the changelog contains "* " and a bullet is used instead.

 In this context PK should be able to always do the right thing though, as that byte is part of the changelog syntax ... dito. date/name/ver and the - on the lines after it.
 The description munging is very different, as you are altering free text. As I pointed out it does the wrong thing for python-peak-util-addons etc.

>> The only way I think this would work would be to write a mail to fedora-devel
>> and ask maintainers to clean up the use of ”, ” and ’.

> Seems reasonable to me.

As I said, those might not be the only chars in question. I suppose "é" will be replaced by "´" right ?

If that's the case, then PK should learn how to display them properly instead of asking people (packagers / translators) to write their own language in an incorrect way (i.e. without accents for non-english languages).

FYI, I'm seeing ü OK with gnome-packagekit-0.4.0-2.fc11.x86_64, though I do get “advanced” instead of ``advanced''.

See http://packagekit.org/pk-faq.html#markup