Re: Xy3 - reference manuals

["Robert Holmgren" holmgren@xxxxxxxx]

** Reply to message from Nathan Sivin  on
Wed, 15 Jul 2009 09:52:42 -0400


> This is a bigger job than it appears. No OCR program is perfectly
> accurate, so someone has to proofread everything. And any document that
> is not clearly printed on clean paper is likely to contain a lot of
> mistakes. I will be glad to pitch in on the conversion part of the job,
> but it will be way too much for one or two people.

It's a HUGE job. I've done it, with several Xy-books, e.g. the
Signature XPL User Guide (http://users.datarealm.com/xywwweb/XPL.PDF),
which was only 156 pages, but if you read the "Editor's
Preface", you get some idea of what is involved (a LOT!). Plus
there are graphical characters embedded in the files (key images
etc) which need to be handled specially, if you want to
reproduce the original correctly.

There are three big issues: What, How, Who.

What: The most complete and useful manual by far is the XyWin
Reference Manual. It covers the same material as Xy3 and Xy4,
but adds new material for Win32, which means it covers 99.9% of
NBWin's capabilities. It combines the two Xy4 manuals in a
single volume. If you're going to do just one, that's the one.
The others are just a waste of time. (NBers would LOVE a real
manual -- they've got nothing.)

How: OCR, duh. Scanning as graphical images is idiotic;
relying on Acrobat Pro or the like to do the OCRing is
unreliable. You need to train the OCR engine to read the font
and the graphics. I use ABBYY FineReader. I've used Caere in
the past. I don't care how fast or good your machine is,
scanning takes a giant chunk of time -- GIANT! You need 300dpi
at least (600dpi will create extremely large TIFFs that take
forever to OCR, but occasionally its necessary anyway because
the typeface is small). You have to un-bind the book (take all
the signatures apart), then make sure that the individual pages
are clean (no dirt or smudges) and perfectly aligned (not
off-axis or "tilted").

Who: Ideally, one person should do it -- or several people with
identical software. Otherwise there will be dissimilarities of
many kinds, finally of quality.

Because this is soooo daunting, I've always hoped to lure an old
XyQuester to contribute their digital text copy of the Manual to
the effort, supposing that they retain an archived copy. The
books were **written with XyWrite**! I am told that TTG's
archives of the XyQuest text files were "lost" by some moron TTG
hired to do computer maintenance. Very few XyQuesters probably
had direct access to this stuff, and those who might still have
the text would probably be worried that somebody holds residual
rights to these manuals. Regrettably, none of the old
XyQuesters are still members of this List (to my knowledge). If
anybody knows any ex-XyQuesters (people like Rose Intingaro, Tom
Atwood, Tim Baehr, Jason Balliet, Lucas Gonze, Ruth Gray,
Susan Koenig, Doug Kramer, or Claire Toupin), please let me know
privately -- or contact them privately.

-----------------------------
Robert Holmgren
holmgren@xxxxxxxx
-----------------------------