[Apb-development] Re: Fw: some new apb features...

[Darron F. <da...@fr...>]

I snipped a bit for readability and will try to answer the questions raised.

> > > 2. Page caching - so that a local copy of page on the other side of
the
> > > bookmark can be sucked down and put into the database for
safe-keeping.
> > > (It's still a little buggy right now but we've got the basics down
pat.)
>
> Wow, you don't know how happy I am to see this.  I've bee wanting to get
around to this feature
> forever.  I'm pretty happy with your implementation.  Mainly I've only
ever been concerned with
> getting the text of a single page with this feature (for archiving faqs,
technical info, etc), so
> making sure images and links work isn't a _big_ concern, but obviously
most people are going to
> want those things to work as well.  The feature has the potential to open
a large can of worms...
> Do we get/store images?  Do we cache on a regular schedule?  Do we
maintain a cached page version
> history?  Do we do any recursive digging?

All images are downloaded and put into the database - if the same image is
used multiple times on a page, only one instance of that graphic is put into
the database. It's a one time download - no scheduled update - we saw it as
a way to freeze that page in time to make sure you could always get to that
content.

We had talked about a cached page version history - keep multiple versions
of those pages around but hadn't really done anything about it yet.

At this time, there isn't a recursive download either - it's just that page
and any directly linked images and stylesheets - basically anything with a
src="" tag.

One thing I forgot to add is that it uses the curl functions in PHP which
depend on the curl library being present. You can grab it here:

http://curl.haxx.se/

> But mainly, what you've done so far, I'd be happy with as a user.  I think
a table at the top of
> the cached page (a la Google's cache) clearly stating that it's cached,
the cache date, and
> explaining that the links may not work is needed.

That is coming - clearly stating that this is a cached page and that the
original is at this url. If you've ever used IE 5 on a Mac you'll notice the
similarities to the Scrapbook feature - I've always wanted a network/browser
independant version of that.

> > > We have some other ideas as well (move to using Pear DB for database
> > > abstraction, complete the multiuser portions, export to browser
bookmarks,
> > > etc.) - but those will come later - we just got slammed with a huge
> > project
> > > that will take up a huge portion of the next month.
>
> I don't know what Pear DB is.  I need to look into that I guess.  We're
kind of MySQL-centric.

PEAR is basically a bunch of code libraries that are distributed with PHP -
it's a way to write DB code without making it MySQL centric.

Here's an example PEAR::DB query:

$sql = "SELECT something FROM somewhere WHERE table = '1'";
$result = $db->query( $sql );
while ($row = $result->fetchRow()) {
    // loop through every row returned and do something
}

With this code - you define your master $db object in a central include file
and can change database types by changing a single variable - it's pretty
sweet. We're pretty MySQL centric here as well but as long as you're using
pretty standard SQL it should work with any PEAR supported DB - Postgres,
Oracle, Sybase, MySQL, etc.

> I want to see the caching.  Mainly I'd like to see the UI for the
authenticated apb user; how they
> cache a page, etc.

Right now, all you have to do to cache a page is flip a radio button when
you're editing or adding a bookmark and it will download it all and insert
it into the DB.

> I think I can safely say that we're interested in adopting both changes,
but there are two issues.
>  The first is mainly UI, and Brandon's in charge of that.  The second is
how to integrate your
> changes into our dev tree.  I don't have a lot of experience at
collaborative-internet-GPL
> programming.  Maybe we need to start thinking about CVS/arc/subversion.
For now we can just
> brute-force it with diff.  If you have any thoughts about that, let me
know.

I'll post the first simple change when I get a chance tonight and after
we've merged our changes with your new version of apb and fixed a couple
small bugs I'll send over a diff - they shouldn't be too hard to isolate.

If you guys want to put up a CVS repository somewhere we're pretty
comfortable with that. Do you have one already?