Yahoo Tuning Groups Ultimate Backup tuning Re: Ascii or Html FAQ

Hi everyone,

Maybe it will help to explain motivation for the database approach very simply.

Also to simplify the concept itself as much as possible, to be easier
to understand and use, if wished.

First, why use a url tree, rather than just send all the original
documents to John to arrange.

Reason is because if you just send the url, you can keep on editing
your document in response to comments, and don't need to keep resending
it, which will save a lot of hassle for everyone.

This was my example of what it would look like, with a few entries
added, enough to show how the tree structure works:

http://homepage.ntlworld.com/robertwalker/tree/tree.htm

Now, why use the database to add an entry, rather than just send the
urls to John one at a time, telling him where you want it to go?

Well, I suppose one could just send them to him, but the database will
be much lower maintenance.

How does it work then?

Idea is that one wants to be able to add an entry anywhere in the tree.

One naturally needs the url which in the case of a simple ascii entry will
point to the file you have uploaded into the Files folder.

Then you need a short description to use in the tree itself
when referring to the entry.

Finally, it is useful to have a long description as well. That's not just
so that you can show it in the status bar, but also tells whatever program
is going to organise the database what you want your document to be called.
But,this is optional, and could be left out if thought too complicated.

Since the tree could have several folders with the same name, one has two
choices = to give the entire path up to your entry, or just give the
folder, but ensure each has a unique name.

However if one reflects on it, remembering that the folders could be
nested several levels, the idea of just having to give one folder has
many attractions.

To make sure each has a unique name, I suggested adding the date of creation
of the folder, and showing that in the tree.

So to fill in the folder field, you need the name of the folder, and the
date it was created. To find that, just go to the folder in the tree,
and read it off.

Maybe one could use some other approach that would work as well or better.

Finally, to add a new folder, as one may want to do from time to time,
one needs to give the new folder name, and the name of its parent. As
usual, one also needs to give the date for the parent, and since one
is making a new folder, one will give the current date as its date.

No more info is needed, so you leave the other fields blank, and it will
then be clear that it is a folder.

So far, this will certainly work, and would be very easy to program for.

If all one wants are text or html documents with internal links, and
none link to any of the others, then that's it done, no more to do.

Will make me happy, - hardly anything to do - maybe a few dozen lines
of code plus some cut and paste of code libraries I've already written.

John won't need to do much either until the last minute when he needs
to put them all together to make the final faq.

At that point, my program can ftp all the entries, and just order them
one after another in the order of the tree. At that point he can also move
the folders around to arrange the sections, e.g. using Explorer or
a tree control.

With a mostly text based FAQ, and no cross-linking, there's no need
even to upload all the faq entries until the last minute when
everything is ready.

The reason for all the other complications is that I had in mind that it
would be an html type of tree, in which any page can refer to any other.

E.g. if you happened to be writing about the meantone temperaments,
and felt my meantone calculation page was relevant, you might want
to add a link to it.

But, how to do that.

To make that work, you need a url or other way of cross-referencing.

Start thinking about how to make that work smoothly and you will
perhaps be lead to the same ideas I came to.

Or possibly some other approach. Maybe there is some other way
that would work better.

Finally, if you want to include images and sound, that may require
further thought about how to arrange all the folders.

I suggested keeping all the drafts in their original places, but
also making an automatic copy of them in the prototype web site.

There, the html documents would be saved in a folder using the
same name you give in the Author field.

Ditto for text documents.

E.g. if Margo Schulter does one of her ascii posts on Neo Gothic,
it would go into the margoschulter folder
and to refer to it you would use
http://tuning_faq.com/margoschulter/sumer_is_icumen_in.htm
or whatever it is.

To make this work, we need a second copy of the tree that
links to the files on the prototype web site instead of the
original drafts.

My program would then automatically ftp all the original files
into this folder, place appropriate links into the tree control
and also add the html navigation links at the top of each page.

Here was my example of what it would look like:

http://homepage.ntlworld.com/robertwalker/site_ex/tree.htm

I did it by hand, not having written the program yet, but it is
all totally routine and easy to do and program for.

Now, because the tree control only works within frames, I added
extra links to the top of the page to hide the tree.

So when you want to link to a page, first navigate to it in the
tree, then click hide tree. You will then see the url proper,
and be able to copy it to use in your link.

If writing in ascii, that's okay - just paste the
link into your document.

My program can then save your document as html with the
these all made into links.

Right, so far we've got an html site for the FAQ. This will
work fine for Newbies who like to use html browsers.

It can even include music, and graphics, and vrml models too,
as would be totally appropriate for an html site.

Here, I added the bit about need for html authors to
specify sub-folders to put their graphics and music
files into, to avoid getting them muddled with
each other.

The method I suggested there was a bit more complicated than
it need be. Easiest is just to add the folder name to the
author field of the table

So database entry then would be something like this:
CPs Sets in VRML <date>; dekany ; http://..../dekany.htm ; robertwalker/dekany

That would put my dekany.htm file into a new folder called dekany
in my main folder on the faq site. All the midi clips referred
to in it would go into the same site.

Okay, that's just a subtlety to do with how the html version of the
tree will work.

NOw back to the ascii.

What's going to happen to all the cross-links when the tree
(or part of it) is exported as an ascii faq?

I suggest that they are just converted into references to the
authors and titles of the documents.

E.g.

http://tuning_faq.com/margoschulter/sumer_canon.htm

would be converted to

See Margo Schulter's article "Sumer canon, discant cadences, meantone"

Graham Breed raised the point of what to do about ascii diagrams
and normal graphics.

One could make a document that incorporates both, with text
instructions within the document to say which is which.

But on reflection, and to stop it getting overcomplicated,
it's probably better if one adds an extra field to the database
saying

Suitable for Ascii (Y or N)

Then when it is time to make the ascii FAQ, only entries marked
as suitable for ascii will be used.

If an entry isn't suitable, but could be converted, the author
could be asked if they are interested in converting it.

Graham Breed also raised the question of what to do about
links to bookmarks in html, and to references to sections in ascii.

For section references in ascii, easiest is just to leave them as
they are.

There is no way a program can figure out that the text

"see 1.1.2"

is a hyperlink to section 1.1.2 when it is all in ascii, and everyone
is using slightly different formats (and the page is perhaps full
of numbers, decimal points, and so on).

This is the sort of thing a human can convert easily , but takes
massive amounts of programming - easier to convert the whole faq
by hand. I'm not volunteering to do it!!

It could be done easily however if one used some kind of conventional
tag, say $$ or something, to mark the sections, and the references to
them, so the program knows what they are all about, with the tag removed
from the final ascii of course.

The other way round, the program can easily find all the hypertext
links in the html, and convert them to ascii section headings, and
1.1.1, 1.1.2, ..., and "see 1.1.2", "see 1.1.3", ...

Problem then is how to get it to choose a sensible numbering system
to fit in with the layoaut of the page, and again, this is
for all practical purposes impossible, but could be made easy
by devising some suitable tags.

If that's too complicated, then the only option is to leave out
all the links.

I reckon that maintaining html documents that are also suitable for
ascii could be pretty tricky to do.

But, many html documents don't really use internal links that much.

E.g, a common format is to list links to all the section headings
at the head of the page, and then at the end of each section, have
a link that goes back to the top of the page.

That would work fine in ascii, as one can't hyper-jump from one
point to another anyway.

Just leave the list of section headings at the top, but change
them into plain ascii.

Maybe it won't look quite as nice as if it was originally
composed in ascii, but will be perfectly readable, and
if there is anything that seriously impairs readability,
John can edit it (or get someone to do it, considering the
likely size of the FAQ).

And, there we are, I think that would work.

Any comments?

Robert

Re: Ascii or Html FAQ

🔗Robert Walker <robert_walker@rcwalker.freeserve.co.uk>

🔗JSZANTO@ADNC.COM

🔗jpehrson@rcn.com