so I try to explain a bit more detailed.
Søren Hauberg schrieb:
> Hi Thomas,
> Thanks for the input, but I'll be honest and say that I don't
> understand it well enough. Neither David nor myself really know HTML/CSS
> so a lot of the web page is made from trial-and-error experiments.
> Questions to your comments are within...
Trial-and-error is the only way (so that is what I've learned) the only way that
html programming can be done. You can validate of course your page with the w3c
validator, but do the ie, moz, opera and (don't forget about this one) lynx all
display the same: NO. So what else can you do instead of trial-error?
> Thomas Treichl skrev:
>> - You could give a robot the chance to find out what each of the pages
>> of the
>> octave-forge website is about, adding the following fields in the
>> <HEAD> of
>> every html file (especially for the packages to make searching
>> available from
>> a search engine)
>> <!-- General meta-types, sorted by their names -->
>> <meta name="author" content=""/>
>> <meta name="date" content="2005-08-01T08:00:00+00:00"/>
>> <meta name="description" content="What ever"/>
>> <meta name="keywords" lang="en" content="Octave-Forge"/>
>> Details about these meta tags can be found at the w3c.org.
> So how does information like this help robots (I'm asking because I
> really don't know)? And how does it help us? The problem with adding
> this information is that the pages are being auto-generated, so the
> information also needs to be auto-generated. So if we should add these
> tags we need to do it automatically.
I have to start somewhere else, before explaining the robot.txt part. It all
The main reason is that Internet became too large already some years ago that
search engines are able to scan every site and every file that can be found. A
typical spent time of a robot (some are also known as crawlers) is about 2..10
seconds every week (okay it depends on the number of documents you have on your
site, for a electronic biblio it must spent much more) checking for changes. So
let us play the game "if I were an search engine", how would I find out what
octave-forge can do for me.
First I would visit the dmoz.org Open Directory Network and would build a basic
database with the information dmoz has got for me. It is necessary that
octave-forge is listed there because I#d say 99% of all search engines get there
basic informations about internet sites from there, ie. names of internet sites.
After that I know that there is an internet site called octave.sourceforge.net.
So me as a search engine go there to the octave.sourceforge.net site and my
entry point of looking around isn't the index.html file but more the robot.txt
file. The robot.txt file tells me which directories are interesting for me and
which aren't, tells me further which directories are allowed for me and which
aren't. I scan directory over directory for *html files (we already know that
also *pdf files are scanned and *doc etc.).
But what about each *html file, counting words to know what the file is about?
No I first look in these meta tags that give me the information (like in a paper
publication) who is the author of the page, when was it last modified (so I can
update my own database), what is the describtion of the page and what are the
keywords that are very interesting for me.
So me as a user, sitting in front of my mozilla, enter eg. the keywords "octave
communication". The best think that could happen now is that my 1st search
result is the communications package of octave-forge. I click this link and get
directly to David's package...
>> - Add an extra <br> on the news page between every news paragraph with
>> date stamp. The news appear to be upset.
> I think this should be done with CSS instead of a <br>. From what I
> understand w3 is advising against using <br> for layout. But thanks for
> pointing out the problem.
However you programmed that, I just wanted to say to add an extra line break there.
>> - Don't forget about the robots.txt file in the root directory of
> Again, I'll show my complete ignorance: what should that file contain?
>> If I have some other ideas then I will contact,
> Please do :-)
same scripts as I can see) into the HEAD of every *html file. You could do
somewhere. Then just add the following line in the html file
<script src="../<relativ path from html file>/javascpt.js"
once from the user's browser, because the browser knows that it has loaded the
javascpt.js file on the last octave-forge website that was visited. It's like
loading and removing a DLD in octave ;o)
Another tip is (needed for older browsers) add the favorite icon describtion
also in the HEAD of every html file
<link rel="shortcut icon" href="../favicon.ico" />
That's it from my side,
Thanks for contacting,
>> Nevertheless, thank you for the great website,
> Thanks. As I said earlier none of us really know what we're doing, so if
> you have HTML/CSS knowledge and want to help out, it would be great.