Re: OODL: Log

[DeRobertis]

At 12:53 PM +1000 on 2/1/00, Paul Sutton wrote:

>Adrian: Any old search will do.  All subjects are prefixed with OODL:

The search engine will not search in the subject field -- it ignores
any text in the subject field when searching. No idea why.

>Adrian: But I'd have to download the entire web archive (plus adds and
>other crap) every time I updated the Log (I can't tell what's new
>remember).  That's simply not viable.

Nope -- only the index pages, which can be done by date, and are not
that big. Besides, the old index pages never need re-downloading.


>Adrian: Let me put it this way: if I were to send you an email saying
>"Anthony, I can't get interpreter to work.  It just gives heaps of
>errors.  Please fix it."  Would you be able to fix it?  I doubt it.

If you gave me a program which looks at the Interpreter sources and
spits out the line number and file where the error is and even a nice
error message, yes. The W3C validator does that.

Here you go. The error reports. Enjoy.
<http://validator.w3.org/check?uri=http://ufp.uqam.ca/OpenCard/Log/>
<http://validator.w3.org/check?uri=http://ufp.uqam.ca/OpenCard/Log/type-frame.ht
ml&weblint=&pw=>
<http://validator.w3.org/check?uri=http://ufp.uqam.ca/OpenCard/Log/allevents-fra
me.html>
<http://validator.w3.org/check?uri=http://ufp.uqam.ca/OpenCard/Log/definitions.h
tml>

It does display correctly. However, I don't like HTML errors. And there
is no guarantee that it will display correctly on any browser --
coments about shooting browsers aside.

>Adrian: Unfortunately the HTML standard keeps changing making life rather
>difficult.

HTML allows you to specify which standard you are using. Put a doctype
up top, and be done with it.

>And many of the changes make newer version incompatible with
>older versions. (The <BR> or <BR></BR> thing for example).

The <BR> has never had -- and probably never will have -- a close tag.
The <FRAME> tag always has had -- and always has required -- a close
tag. And so on.

>You and Uli
>have mentioned a number of times that there are few if any ANSI C++
>compilers that are completely compliant, the same problem exists with web
>browsers.  All I can do is write simple, generic HTML and fix problems
>when I'm told about them.

Why not write HTMl that conforms to the standard? Especially if a tool
is generating it?

Uli and I do note the lack of compliancei n compilers, and while it
does stop us from fully using certain things, such as templates, it
does not stop of from making the rest of the program complaint to the
standard -- or as best as possible. We don't use NULL as a place for a
quick temp, although on MacOS we could get away with it. We don't
depend on the size of a long or a int or a short (or at least try not
to). Why? Because we have no guarantee that it will work everywhere; in
fact, I can name systems where all of those asumptions will brake.

[I know the old interpreter was pretty bad -- the new one will be much better]