The resource bundle and resource md5 sums.

[Wolthera]

Okay, so this is pushed.
http://commits.kde.org/calligra/a56da8b3c3ddc868e937f5fbf39c271d0d98d2f5

Also, to fix a different bug where the preset docker would get confused
about which preset was the one you needed, the new or the old, I put in a
check to have resources not be added if they have the same name as an
existing resource.

This does not use MD5sum, because the resource dockers don't identify with
MD5sums either, but it's easily changed to check for MD5sum instead of name
when that time comes.

On Tue, Dec 23, 2014 at 6:02 PM, Wolthera <griffinvalley at gmail.com> wrote:

> Sorry, it doesn't install all bundles, it only installs activated bundles,
> so only the activated bundles take up start-up time :)
>
> I'll start implementing this then.
>
> On Tue, Dec 23, 2014 at 5:50 PM, Boudewijn Rempt <boud at valdyas.org> wrote:
>
>> On Tue, 23 Dec 2014, Wolthera wrote:
>>
>>  So, right now, the resource bundle uses md5 sums to figure out which
>>> preset is part of it. However,
>>> these md5sums, which are stored in the manifest.xml are bogus.
>>>
>>
>> Hm... But they are the only way to figure out whether two resources are
>> the same. If the md5sums don't work, we've got a bigger problem!
>>
>> Right now, the actual details have gone quite fuzzy in my mind, and I'm
>> trying to focus on fixing the shortcuts/menu mess, so it's kind of hard to
>> figure out what's up.
>>
>>  I do not know why they are bogus, but I am hoping to implement a bit of
>>> functionality that is half
>>> work-around and half kinda necessary anyway:
>>>
>>> First, Krita always installs all bundles upon start-up.
>>>
>>
>> That's actually something that needs fixing then -- because one of the
>> nice things of the bundles is that blacklisted/uninstalled bundles
>> shouldn't take memory or startup time.
>>
>>  > I suggest we us this chance for the bundle
>>
>>> to collect all the md5sums of the resources it installs.
>>>
>>> Then, we get it to check it against all the manifest md5sums, when these
>>> are mismatched, we tell the
>>> user "Ah, it seems that these brushes have mismatching md5sums with the
>>> bundle's manifest. They may
>>> have gotten corrupted udring download:
>>> <list of brushes>"
>>>
>>> That way, the user actually can see the use of these weird numbery
>>> things.
>>>
>>> Then, when Krita deinstalls a bundle, it'll use the md5sums of what it
>>> knows it installed instead of
>>> the list in the manifest.
>>>
>>
>> We can do that, yes.
>>
>>
>>> Does this sound like a plan? We can then later figure out why the
>>> manifest md5sums are nonsense.
>>>
>>> --
>>> Wolthera
>>>
>>>
>>>  _______________________________________________
>> Krita mailing list
>> kimageshop at kde.org
>> https://mail.kde.org/mailman/listinfo/kimageshop
>>
>
>
>
> --
> Wolthera
>



-- 
Wolthera
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kimageshop/attachments/20141223/b0eaffc6/attachment.html>