[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: [Bug 154698] Won't open some TIFF files
From:       Andreas Pakulat <apaku () gmx ! de>
Date:       2007-12-28 1:35:05
Message-ID: 20071228013505.GB3715 () morpheus ! apaku ! dnsalias ! org
[Download RAW message or body]

On 27.12.07 18:00:51, James Richard Tyrer wrote:
> Andreas Pakulat wrote:
> > On 27.12.07 16:47:53, James Richard Tyrer wrote:
> >> Pino Toscano wrote:
> >>>>> Okular just uses what KMimeTypes reports to be the mimetype of 
> >>>>> the file, and then uses the backend responsible for handling that
> >>>>>  mimetype. If the backend receives something it cannot read, it's
> >>>>>  not its fault.
> >>>> That is not actually the case.
> >>> It's nice to see how you reject the explanation of the developers of 
> >>> that architecture.
> >> Perhaps if you read the rest of what I said, you would not jump to
> >> conclusions.
> > 
> > And perhaps you could stick to discuss this on the bugreport and not on
> > a bunch of mailinglists?
> 
> First, the okular-devel mailing list is automatically CCed if I put 
> something on an Okular bug report

But its not an okular bugreport anymore, the only CC for the bugreport
is the kdelibs-bugs list which you didn't send to.

> so there isn't a "bunch of mailing 
> lists" there is ONLY ONE (I never considered two to be a bunch and I 
> don't think that M-W does either -- but I am certain that one mailing 
> list isn't a bunch).

Whoever M-W might be, but I hope you agree that also Cc'ing Pino is
quite a bit too much as you can be quite sure that he reads the okular
list, can't you?

>  Second, I am following up to a previous posting to 
> the kde-devel list.

Then why don't you reply properly in that thread?

> Third, this appears to be a wider issue which I would like to discuss
> with some other developers

Then you should 

a) open a new thread
b) use the right mailinglist, stuff like KMimeType is discussed on
kde-core-devel.

> -- developers other than PT who doesn't appear to understand what I
> said and then tells me that I am wrong based on his misunderstanding
> of what I said.

So far I didn't see anything wrong from him. In fact he quite well
understood why okular behaves as it does currently.

> >>> Okular just trusts what KMimeType reports as results. No less and no 
> >>> more.
> >> Probably not the greatest of ideas since KMimeType is known to be
> >> somewhat broken in KDE3 and I doubt that it has been fixed.  Actually,
> >> it needs some design work first.
> > 
> > You've got a bugreport number?
> 
> Well there would be more than ONE bug report.  Probably quite a few.  I 
> have one that I filed (I may have filed more):
> 
> http://bugs.kde.org/show_bug.cgi?id=132673

Thanks.

> >> It is here I can demonstrate what I mean when I say that KMimeType is
> >> somewhat broken.
> >>
> >> Make a copy of the first attachment and remove the '.tif' extension --
> >> have NO extension.
> >>
> >> Now try:
> >>
> >> 	kfile4 --dialog <file>
> >>
> >> with the file with no extension.  Notice that you now have a preview.
> >>
> >> Do the same from a Konsole with Okular and you will find that it opens
> >> correctly.
> > 
> > Which is another indication that KMimeType is the "problem". Though the
> > problem here is a quite well working optimization. You really don't want
> > KMimeType to completely read all files that are thrown at it, if it can
> > guess the type from the file extension.
> 
> Yes, however most KDE3 applications don't have the problem and GwenView 
> 2.0 for KDE4 doesn't have an issue either.  I haven't yet dug into the 
> code either so I don't know what they do differently -- is it really the 
> job of a bug reporter to have to dig into the code to get a bug accepted 
> as real? 

No, but you need to trust the original authors of a software to
understand their own software and know how/where a fix for such a bug
needs to be applied.

> Obviously, GwenView does something slightly different than the 
> Okular front end and *therefore* it doesn't have the problem.

I didn't read any code either, but I suspect that Gwenview simply reads
the file and determines the format of the binary data from that. More
about that in my other answer.

> >>> Thus, I'm reassigning the problem to kdelibs (where KMimeType is), as
> >>>  should be KMimeType to report us the correct mimetype for it.
> >>>
> >> As I see it, there are issues with BOTH Okular and the KDE MIME type system.
> > 
> > This is a probably a wontfix, unless you want to introduce a lot of
> > extra IO for very little gain into KMimeType.
> 
> Not knowing the code, I couldn't say how much would be needed. 
> Remember, it doesn't work correctly at the present and this really does 
> need to work correctly.

Why? I mean, sure it would be better if a misnamed file could still be
shown properly by okular, but I don't see this as a severe issue. In
fact I find it quite good that an error message is posted, else the
person who did misname the file might send the misnamed file to somebody
who then can't open it at all because he/she doesn't use a software
thats as intelligent as Gwenview.

Andreas

-- 
You are fairminded, just and loving.
 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
[prev in list] [next in list] [prev in thread] [next in thread]