[Clipart] Sea of Subjectivity

[Matthew Gates]

On Sunday 29 Jan 2006 10:16, B Kreps wrote:
> > So what do you propose? To create from scratch a new backend?
> > This was our initial plan but after a long time reached the
> > conclusion that we don't have the resources (people, time,
> > skills) do to it and the best chance is to reuse something which
> > is already existing, working and under active maintenance.
>
> Well, I'm entirely new to this list and I haven't looked at the
> backend yet, so I can't comment on actual implementation.
>
> What resources are available? What is the ultimate goal and
> timeline? Given that there are already 5k or 6k images already
> online, it might be best to avoid a solution that requires
> processing all those submissions again - if it is even possible.

It shouldn't be too bad to re-process these files if it is necessary.  
It's not so much data for a modern machine and a half well written 
program.  If there's a way to avoid it OK, but we shouldn't sweat it.

Having said that, my SVG -> PNG conveter using gimp seems to eat all 
my memory and cause my machine to freak out a little (I'm still 
looking into this one - perhaps there are some files that cause Gimp 
to freak).  However I think processing XML data -> XML data is a 
little different - much lighter.

> It would be good to find out what people think of implementing a
> "collection" format and submission process into the web forms. It
> is already possible to submit an archive file, i.e. tar.gz, zip,
> etc.; so if a collection archive AND a collection sample could be
> submitted at the same time, then the sample could be viewed in the
> main library, and an additional informational field would designate
> this sample as representative of a collection (i.e. a deck of
> cards) to the user. The archive could then be downloaded. A system
> like ccHost could then be implemented on top of a combination of
> individual clips and archives of collections of clips.
>
> Thoughts? Maybe this needs to be a new thread in the list.

My main beef with the current setup is that there is a lot of 
duplication between categories.  In v0.18, almost 1000 files are 
duplicated, some as many as 6 times, making a total of over 1000 
unnecessary copies.  It's quite a big problem IMO - I believe it 
should be a goal to minimise the size of the download - both for the 
sake of saving bandwidth/load on the server, and for the convenience 
of users.

If we're going to start a re-design, this needs to be thought about.  
Something more like KimDaBa's organisation where all the images go 
into a big repository and flags assigned to each file allow quick 
searching.  Of course this needs some sort of application.  More 
techie Linux users would probably be quite OK with an index file and 
grep, but windows and Mac users might not be so accepting of this 
sort of thing.   ;-)

Perhaps it's possible to implement something that works in any a web 
browser?  Just an HTML file distributed with the archive? Javascript 
or something? Simplicity and portability is the key IMO.
-- 
Matthew Gates
http://porpoisehead.net