[PhotoFloat] Spring cleaning (video, preloading, threading, updated JS-libraries, etc)

[Tom Talpey]

On 6/4/2015 2:36 PM, Joachim Tingvold wrote:
> On 4 Jun 2015, at 15:46, Tom Talpey wrote:
>> Looks like the scanner can't be interrupted with ^C with these patches.
>> The first interrupt throws an exception and scanning stops, but the
>> script doesn't exit. Subsequent interrupts throw another, but again, no
>> exit.
>
> Yes, that was a side effect when doing threaded thumbing. I never really
> bothered to look into it since I never use ^C (-:

Ok, I'll take a look when I have a chance. I find that I often get
errors during scans (missing metadata or permissions errors) so I
frequently interrupt to fix problems as it goes.

>> Any chance you might look into implementing a file structure on the
>> cache?
>
> This is on the top of my "things I'll fix whenever I have the mood to do
> it"-list.

Excellent. I don't remember whether this was something Jason said he
would consider, but I'm glad to hear you agree.

> I also want to look into dedupe -- I currently have two galleries; one
> primary that everyone can see, mostly consisting of event/time-limited
> events (i.e. different albums), and one secondary, containing /all/ of
> my photos (sorted into <category> / <year> / <month> / <day>) that is
> password-protected (BasicAuth) so that I can view all of my photos (even
> if they're not part of an album, which is practical f.ex. for mobile
> photos). The cache-folder for my primary gallery is like ~10G and around
> ~20G for the secondary. Since all pictures are part of the ~20G, I
> shouldn't really need to use those ~10G for the primary gallery.

Could this be done with a filesystem utility? There are some tools
already out there to clean up and dedup filesystems by calculating
hashes and making links. Since PhotoFloat doesn't rewrite anything,
hardlinks would be safe and it seems pretty easy to adapt them as
a post-processing step of scanner.py?