Re: KDE/Qt Flatpak SDK/Runtime

[Brüns , Stefan]

On Dienstag, 15. November 2016 18:00:34 CET Aleix Pol wrote:
> On Tue, Nov 15, 2016 at 1:16 PM, Matthias Kuhn  wrote:
> 
> I would like to see how much pyqt increases the size of the (not yet
> optimized) runtime. I agree it can be interesting, but I'm not sure by
> how much.
> Also the fact that there's so many approaches for python Qt bindings
> doesn't make things much simpler, and that's python alone. Does that
> mean we want to have all languages in there? I'll be happy to work
> with anyone who wants to get PyQt in their application, if we see that
> we're getting so much projects that need PyQt that it makes sense,
> then by all means it should go in.
> 
> Having different run-times for applications that use KDE Frameworks
> and applications that don't feels like shooting ourselves in the foot,
> at this moment. Note that in Flatpak runtimes don't extend each other,
> they duplicate, AFAIK.

If I understand http://flatpak.org/developer.html correctly:
- optional/add-on parts can be referenced using Extensions
- you may use several extensions from application and/or runtime

Also runtimes seem to stack, see e.g. 
https://git.gnome.org/browse/gnome-sdk-images/tree/org.gnome.Sdk.json.in

As far as I understand the OSTree repository format, even distinct runtimes 
which have overlapping content will not duplicate storage or download 
requirements, as OSTree is a content-addressed storage, although I may be 
wrong here.

Kind regards,

Stefan

Re: Review Request 113587: Add support for x-gvfs style options in fstab

[Brüns , Stefan]

On Montag, 13. März 2017 22:37:51 CET Albert Astals Cid wrote:
> > On March 13, 2017, 2:29 p.m., Kai Uwe Broulik wrote:
> > > Sorry for the long wait… I was under the impression it landed a long
> > > time ago.
> > > 
> > > Itook your patch and ported it to Frameworks 5, please see the review at
> > > https://phabricator.kde.org/D5034

Would be nice if you retained the original copyright.

I have no phabricator account (and there is no method to register, AFAICT), 
which bars me from adding any comments there ...

Kind regards,

Stefan

Re: Review Request 113587: Add support for x-gvfs style options in fstab

[Brüns , Stefan]

On Montag, 13. März 2017 22:37:51 CET Albert Astals Cid wrote:
> > On March 13, 2017, 2:29 p.m., Kai Uwe Broulik wrote:
> > > Sorry for the long wait… I was under the impression it landed a long
> > > time ago.
> > > 
> > > Itook your patch and ported it to Frameworks 5, please see the review at
> > > https://phabricator.kde.org/D5034

Would be nice if you retained the original copyright.

I have no phabricator account (and there is no method to register, AFAICT), 
which bars me from adding any comments there ...

Kind regards,

Stefan

Re: releaseme new requirement: non-conflicting files on case-insensitive FS/OS

[Brüns , Stefan]

On Montag, 20. November 2017 18:50:34 CET Reindl Harald wrote:
> Am 20.11.2017 um 18:30 schrieb Thiago Macieira:
> > On segunda-feira, 20 de novembro de 2017 07:58:06 PST Reindl Harald wrote:
> >> hopefully this means also that filenames in the future don't contain
> >> uppercase chars
> > 
> > Why would we want that?
> 
> well, i don't know many projects or people which create binaries other
> than lowercase chars and using only lowercase filenames would avoid the
> whole topic at it's root
> 
> to be honest: i am annyoed that "lsof | grep DEL | grep /usr" brings

OUCH!

Fixed that for you:
lsof | grep -E "\sDEL\s.*\s/usr"

Regards,

Stefan

Re: Is it possible to mock an external drive (for unit testing)?

[Brüns , Stefan]

On Dienstag, 13. März 2018 15:38:54 CET Michael Heidelbach wrote:
> Thanks for your reply, Roman
> I also thought of archivemount but (maybe to early) discarded that idea
> because for testing I need control over the device id.
> I'm not sure how to make identification more reliable either. If I'm
> able to read the uuid of a drive or partition it would be small enough
> to incorporate into the document id directly. Otherwise I will have to
> do some heuristics and use an extra db for device id/drive mapping. I
> really don't know yet.
> Cheers,
> Michael

How about using one db per partition? This has several benefits:
- you free up space for 64 bit inodes (currently 32bit inode || 32 bit device)
- UUIDs are large in general (128 bit), adding this to the document id means 
considerably larger keys to compare.
- the db becomes movable, i.e. saved on external drives
- as a consequence, the db can also be saved inside a vault

The partition UUID could be saved inside the db, but IMHO this would be an 
optional safety measure only.

Kind regards,

Stefan

Re: baloo needs to be notified of fuse mounts.

[Brüns , Stefan]

On Dienstag, 20. März 2018 11:45:30 CET Michael Heidelbach wrote:
> Hi!
> 
> In order to fix bug 390830  (== baloo
> should not index plasma-vaults) I need a notification about mounts and
> unmounts of fuse.encfs filesystems (and maybe others).
> 
> I've tried to find something suitable within Solid - without success so
> far. Can you give me some hints, please?

solid does not currently list any fuse mounts. It should be fairly trivial to 
add fuse to the solid fstab module (which, despite its name, also listens to 
mtab resp. /proc/mount changes).

Kind regards,

Stefan