[rescue] miss me?

[Joshua D Boyd]

On Thu, Aug 29, 2002 at 02:05:21PM -0400, Tim H. wrote:

> Perhaps a bit more color?  Like
> > maybe make the background of the current tabbed card a light blue, and
> > the background of the non-current tabs a slightly darker blue
> 
> thought about that, hadn't picked colors yet. It would be a problem,
> because the curved corners of the tabs are transparent, so any
> background color is going to have a square background sticking out.  I
> didn't want to try matching a picture color to the page background,
> because some browser or other will screw it up, even using a web
> pallette, but if I kept the page background white, any browser should
> get that right.

If someone uses a web browser that results in the picture versus the
background being a slight shade off, let the suffer.  Just make sure
that the one you use doesn't, and that preferably Mozilla doesn't.
Also, I find gifs better than jpgs when it comes to color matching
problems. 
 
> I already have a real ugly page I used to populate the database for
> testing, that uses sessions so you don't have to keep entering your
> password.  In normal use I don't expect to be adding multiple entries
> very often.

I tended to find that I wanted to add multiple links at once, but that
is because I tend to queue them a little bit before I get to it.
 
> That is a good idea, A drop down defaulting to current category.  I know
> I have to whack out some javascript to get the user list to default to
> current user anyway.

See, now we are cooking with gas!

> And do you allow people to enter a URL without a name, then have
> > the program see there is no name, and download the URL, and extract
> > the<name field>?
> 
> well, I was thinking about it, but the overhead on that would be a
> little high, since this isn't a real program, it's just a PHP script.  I
> also thought about defaulting the url to the http referer, so if you
> were on a page you wanted to bookmark you could just type
> "pellucidar.net/weblinks" in the address bar, and the page you were on
> would come up in the URL field.  That would actually be easier than
> extracting the page title for the name field.  In fact, off the top of
> my head I can't tell you how I would get the title using PHP.  I'd have
> to get the whole target page loaded in as a variable somehow, then parse
> for the <title></title> tag.  I could do it externally with wget, but I
> don't currently have shell access on the server.

What I did was flag entries made like that, then my home computer
downloaded those entries, downloaded those pages, then parsed the titles
and uploaded them again.  For the <title></title> bit, I assumed that
both would be in one line, and only scanned line by line.  I did it in
perl since it was faster than C, and I didn't know python then.  I
didn't have shell access on the server.  That is why I didn't just have
it add it to a background queue, but instead queued them all to do
nightly.  If you did have better access on the server, loading the whole
page to one variable, would probably be better.  And you could also run
a few background threads to work on clearing the queue of pages that
need the title to be found.  There is all sorts of interesting stuff
that can be done here.

-- 
Joshua D. Boyd