Re: [GMG-Devel] Tags/Collections for batchaddmedia

[Christopher Allan Webber]

Putting it on a Trac ticket would be a great idea :)

Dan Krol writes:

> Should I take discussion about this feature to a trac ticket to stop
> spamming the other goblins?
>
> On Mon, Nov 9, 2015 at 10:14 PM, Dan Krol <address@hidden> wrote:
>
>> Yep, I do have an account on your trac, and looks like it still works.
>>
>> On Mon, Nov 9, 2015 at 9:35 AM, Christopher Allan Webber <
>> address@hidden> wrote:
>>
>>> So I just merged this one, since it looks like and I trust you and know
>>> this was already broken in master, and the code looks good.  Thank you! :)
>>>
>>> For future patches: do you have a username on the tracker?  If so, file
>>> a bug and target for 0.8.1, link the branch, and mark "review".
>>>
>>> If not, I realize the current issue tracker situation is
>>> FU[NQBA(Not Quite But Almost)]BAR'ed.  If you don't have an account,
>>> just let me know and I'll mark in my Orgmode, or you can hop in IRC and
>>> try to convince someone to wait 20 minutes to allow the page to load
>>> which allows us to add new users in Trac currently... ;X
>>>
>>> Dan Krol writes:
>>>
>>> > How about this for a starter commit:
>>> >
>>> >
>>> https://github.com/orblivion/mediagoblin/commit/17eb7786b68c528085fb93f0d6bfd289153e485e
>>> >
>>> > The "add collection" view was broken on master. It was relevant to the
>>> main
>>> > patch I'm working on. How do I go about submitting the patch? Or is the
>>> > link sufficient?
>>> >
>>> > On Sun, Nov 8, 2015 at 2:10 PM, Dan Krol <address@hidden> wrote:
>>> >
>>> >> I'm a bit delayed because I wanted to successfully upload my latest
>>> >> vacation, which required video transcoding, which requires Debian
>>> Jessie,
>>> >> which required something something yak hair.
>>> >>
>>> >> Anyway that's all done; I will try my best to get the patch done
>>> tonight
>>> >> after my other Sunday to-dos. However one edge case came to mind as I
>>> had
>>> >> the transcoding errors, and I'm wondering about your opinion.
>>> Supposing a
>>> >> user were to try to batchaddmedia a cvs's worth of stuff, and come
>>> upon an
>>> >> error. They fix it (maybe they just had to apt-get more gstreamer
>>> stuff).
>>> >> At that point, they shouldn't just re-run the command, they should
>>> edit the
>>> >> csv to remove the items they already successfully uploaded (that's
>>> what I
>>> >> did). Maybe I could just put a warning message to do so if there's an
>>> error?
>>> >>
>>> >> What would be worse, though, is if they chose to use celery to process
>>> >> things. Processing errors (video or otherwise) would be silent, and the
>>> >> other files would proceed anyway. Not such a big deal in general, but
>>> with
>>> >> collections it's an issue because order tends to matter (from a user
>>> >> perspective). They can't just fix the processing problem, go back, and
>>> >> re-upload the ones they missed. Even if we added re-ordering features,
>>> it
>>> >> loses some of the benefit of "bulk" to have to move things around
>>> manually
>>> >> after the fact. So, perhaps I should disable, or at least give a y/n
>>> >> warning, for batchaddmedia with celery, if there are any collections
>>> >> specified in the file?
>>> >>
>>> >> Thanks.
>>> >>
>>> >> On Fri, Nov 6, 2015 at 6:45 AM, Dan Krol <address@hidden> wrote:
>>> >>
>>> >>> Oh that's right I can just make a temporary hacking installation.
>>> I'll do
>>> >>> that.
>>> >>>
>>> >>> On Fri, Nov 6, 2015 at 5:30 AM, Christopher Allan Webber <
>>> >>> address@hidden> wrote:
>>> >>>
>>> >>>> It's prooooobably safe :)
>>> >>>>
>>> >>>> Jessica just landed some major database changes (and now is taking a
>>> >>>> month long break to avoid total burnout, it was a big task).
>>> >>>>
>>> >>>> As a warning, there are some failing tests that I need to address,
>>> which
>>> >>>> I just found yesterday.
>>> >>>>
>>> >>>> It depends how critical things currently are.  You may wish to either
>>> >>>> copy the database over for local hacking and testing and delay your
>>> main
>>> >>>> update... or if some bugs are okay, go ahead and upgrade, and make a
>>> >>>> backup! :)
>>> >>>>
>>> >>>> I'm going to try to get all tests passing again early next week.
>>> >>>>  - Chris
>>> >>>>
>>> >>>> Dan Krol writes:
>>> >>>>
>>> >>>> > Hopefully this weekend. Should I be safe to upgrade to the current
>>> >>>> HEAD of
>>> >>>> > master and migrate my database? (I make backups between upgrades
>>> >>>> > regardless).
>>> >>>> >
>>> >>>> > On Wed, Nov 4, 2015 at 8:38 AM, Christopher Allan Webber <
>>> >>>> > address@hidden> wrote:
>>> >>>> >
>>> >>>> >> That would be great.  We might release a 0.8.1 very shortly.
>>> maybe we
>>> >>>> >> can make it in for that?
>>> >>>> >>
>>> >>>> >> Of course, that's up to what you have time for!
>>> >>>> >>
>>> >>>> >> Dan Krol writes:
>>> >>>> >>
>>> >>>> >> > Great, thanks. I'm assuming I'd want to make it based on master
>>> >>>> rather
>>> >>>> >> than
>>> >>>> >> > v0.8.0? It looks like the db schema related to collections has
>>> >>>> changed.
>>> >>>> >> >
>>> >>>> >> > On Tue, Nov 3, 2015 at 2:40 PM, Christopher Allan Webber <
>>> >>>> >> > address@hidden> wrote:
>>> >>>> >> >
>>> >>>> >> >> Awesome, thanks for the patch!
>>> >>>> >> >>
>>> >>>> >> >> Dan Krol writes:
>>> >>>> >> >>
>>> >>>> >> >> > Okay, I'm guessing that the reason tags weren't added to
>>> >>>> >> batchaddmedia is
>>> >>>> >> >> > that submit_media takes a string parameter for tag names
>>> rather
>>> >>>> than
>>> >>>> >> the
>>> >>>> >> >> > corresponding tag objects. This means that submit_media would
>>> >>>> likely
>>> >>>> >> be
>>> >>>> >> >> > querying for the same tags over and over again, which is very
>>> >>>> >> >> inefficient.
>>> >>>> >> >> > So, it's maybe worth splitting out a function underneath
>>> >>>> submit_media
>>> >>>> >> >> that
>>> >>>> >> >> > takes the tag objects, which submit_media calls on? Do you
>>> all
>>> >>>> have a
>>> >>>> >> >> > recommended way to do this, that's in line with your code's
>>> >>>> >> >> organization? I
>>> >>>> >> >> > assume just adding tag_objects and collection_objects
>>> parameters
>>> >>>> would
>>> >>>> >> >> > muddy up submit_media's interface.
>>> >>>> >> >>
>>> >>>> >> >> I think that should be possible.  We could just add another
>>> keyword
>>> >>>> >> >> argument to submit_media that takes tag objects.
>>> >>>> >> >>
>>> >>>> >> >> > For my own purposes I can just do it the inefficient way for
>>> >>>> now, if
>>> >>>> >> >> > there's not a great answer for this. Here's what it looks
>>> like
>>> >>>> at the
>>> >>>> >> >> > moment (forked from v0.8.0 fo rnow), and it seems to work:
>>> >>>> >> >> >
>>> >>>> >> >>
>>> >>>> >>
>>> >>>>
>>> https://github.com/orblivion/mediagoblin/compare/853b74cd813b3dea3ac351256b3ce1730664d7f9...batchaddmedia-collection-tag
>>> >>>> >> >>
>>> >>>> >> >> Yay!
>>> >>>> >> >>
>>> >>>> >> >> If you wanted to do the more efficient route, I'd quite happily
>>> >>>> accept a
>>> >>>> >> >> patch/branch and merge it.
>>> >>>> >> >>
>>> >>>> >>
>>> >>>> >>
>>> >>>>
>>> >>>>
>>> >>>
>>> >>
>>>
>>>
>>