REBOL TEXTBASE

[Snip]

A bit naive, perhaps. Many (if not most) of the files on my home machines (about 500,000 files at last count) contain `text' that is `encoded' by the programs that use it. Simple examples are .DOC (Windows Word) or .PS/PDF (Adobe PostScript/Acrobat), Mail Files, Help Files, MySQL data bases, Outlook Contact Lists etc. etc. Even text that isn't `deeply encoded' is often surrounded by a disconcering quantity of HTML markup or some such, and increasingly, these days, basic text is often stored in DBMS systems such as mySQL or Oracle. Sadly, recovering `sensible' text from these files is not an easy task without running the processors that interpret it (Word, Acrobat, GhostScript, Communicator, ...) in each specific case. This is one of the reasons that I am not much of a fan of the programs that store things in these idiosyncratic ways---but since the programs are heavily used one generally has to `go along' with customary usage. Rather than taking the approach suggested in this note, I guess I'd suggest starting with an implementation of something like the much-beloved `HyperCard' that was such a popular part of the Mac World for such a long time. Since I never liked Macs much, and since most of the HyperCard-like Windows systems were _very_ expensive, I never used this much, but it had a real following among a broad set of users, and it strikes me---perhaps naively---as something that might be easy to do, at least in a preliminary way, in REBOL. If such a thing were doable, then it might be possible, in some cases, to invoke processes that would convert other forms of files into `HyperCards' that could then be accessed in-lieu of the information in the documents themselves, thus accomplishing much of what you seem to want to be able to do.

I personally would not want everything on my computer indexed, but only documents that I might use in my writings. Anything else would be undesirable clutter. Documents such as web pages can easily be stripped of their tags using parse. Source code for rebol would certainly work as is. Almost all programs have an export to ascii option. Eudora email files are already in ascii format. Louis

Actually that's not true. Complex HTML markup, particularly with in-line java can be quite difficult to parse,

I suppose most computer code would, but it isn't something that I'd normally use in `my writings'.

Some do. Far, far from `almost all'. And in any case this sort of misses the point, doesn't it? If you have to remember to write out an `ascii version' for the purposes of your indexing, then all of the advantages of `automatic' indexing pretty much disappear.

Yes, but Eudora only handles a small fraction of Net mail. NetScape Communicator and Outlook, which handle far more, I think, both store mail in complex structured files.

Thanks! I would like to see your code. Your web page won't let me in, however, (Forbidden You don't have permission to access /~screbol/reb/sucher/ on this server.) so could you just email it to me off list, please? Louis [louisaturk--eudoramail--com] Louis

Ups, i forgot: you have to go there with the goto-button on the rebol-desktop. there is only a %index.r, no %index.html yet..