The resource bundle and resource md5 sums.

[Timothée Giet]

Le 23/01/2015 12:00, Stefano Bonicatti a écrit :
> @Dmitry:
> 1) This is what i would say too, the problem though is that right now 
> there are resources using XML that could be interpreted in different 
> ways when on different PCs.
> Then there's also the problem of gradients files that when read the 
> values in memory are different from the ones in the files and if with 
> no modification you save the gradient again, you will obtain a file 
> that is different from the first one even if it should be the same 
> (you didn't do any modification so.. this is partially why detecting 
> this modification would make sense).
> To conclude, if we keep resources how are handled now and choose to 
> calculate MD5 on static (deterministic) data, there are parts of the 
> non-static data that from one version to the other version changes, 
> but maybe no static data changes, so you would have the same MD5 for 
> two version that are actually different.
>
> 2) Wolthera told me (and i still didn't checked) that there should be 
> few places if not one where resources can be changed, so if changes 
> comes from one source only, why it should be a problem?
>
> 3) I don't think i completely get what you say here. Right now those 
> "in memory only" resources can be saved as normal gradients in the 
> bundles, so it does create a file from them with their correct names.
> The problem though is that if we have to calculate MD5 on files only 
> (the real static data imho), they don't have it.
>
> @Boud:
> Well yes and no, UUID doesn't really guarantee that the resources 
> aren't equal, it just works that it's statistically improbable that 
> two different users can generate the same UUID for different 
> resources, though since the UUID doesn't actually depend on the file, 
> one can create two differents UUID for the same file, and this is bad :P.
> MD5 is the correct way, though here it's either calculated on the 
> "wrong parts" or the resources are implemented in the "wrong way" (if 
> one wants to guarantee to be able to recognize two equal resources by 
> their content).
>
> @Timothée:
> But what if one doesn't want to overwrite the originals and want to 
> prepare a set of bundles for others, or the like? (But it's not only 
> that, please see point 1 of the answer to Dmitry).
>
In this case, saving new presets first and bundling them afterward makes 
more sense.
Without manually saving the preset, what name and thumbnail should the 
preset use? the same as the original one? So you end up having two 
preset with same name and thumbnail? sounds bad...



>
> _______________________________________________
> Krita mailing list
> kimageshop at kde.org
> https://mail.kde.org/mailman/listinfo/kimageshop

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kimageshop/attachments/20150123/a4129d15/attachment.html>