Ref:  Your note of Mon, 25 Sep 2017 01:12:55 +0100

Melvyn Maltz wrote:
> And while I'm on the subject, does anyone remember that a few
> months ago I wanted the PoOPS ripped apart and reconstructed so
> that every single instruction had a page to itself with links ?

It's not for want of trying.  The problem is the sheer size of
the document and the fact that it is conditionally generated from
a repository containing a much larger set of documents, the
primary purpose of which is for the IBM architects and engineers
to design and implement the architecture, and which have for many
years been written only to be formatted in the existing way as
two-column printed or PDF documents.

Back in December 2012, I discussed improvement possibilities with
Dan Greiner and he sent me an "HTML" exported version of PoO from
which I constructed an IBM internal "proof of concept" static web
version of the whole book (at the zEC12 level) with a separate
page for each topic (except if there were multiple instructions
followed by a common description, it included all of the
instruction-only topics on the same page as the common
description).  One could go straight to a given instruction
either via hyperlinks from tables in Appendix B, or using a
front-end menu system which used a JavaScript program generated
from Appendix B. As a collection of static web pages, it didn't
have any built-in search, but it had the same index as the PDF
version, hyperlinked to the referenced topics.

I had many problems with the generated "HTML", many of which were
not the fault of FrameMaker.  For example, markup that appears
identical on the page can be defined in many different ways in
the source, using combinations of named styles, or direct
formatting instructions such as underscore.  Also, the "HTML" did
not include any indication of which borders were present or
absent on a table cell, and in some cases the source used complex
patterns of table cells with selected borders present to draw
diagrams. It was therefore necessary to use "intelligent" logic
to try to deduce which borders should be present from the
context, which was exceedingly messy in places, and provably
impossible in others.

It took me several weeks to get to what I considered an
acceptable quality for usability, at least for a "proof of
concept". And much of the work consisted of writing a program (in
REXX and CMS Pipelines) that made many hundreds of specific
adjustments to the source, where many of those adjustments would
have to be reworked for any updated version.

The "proof of concept" was totally successful as a demonstration
of the level of usability that could be achieved, and even now,
five years later, it is still a useful IBM internal resource.
However, it also demonstrated that the amount of work that would
be needed to convert the book to a professional-quality web
format was enormous, and would probably require migrating much of
the architecture documentation repository to a new format (and
retraining all the users of it).

It would certainly not be at all feasible at present to provide
an officially supported web version (correcting formatting errors
and building new versions promptly for each new release of the
PDF version). However, I personally felt that the "proof of
concept" web version was a very useful resource, even if far from
perfect, so I am continuing to investigate possibilities for
making something similar available "as is", for example via a
Tools and Toys site. Management might be uneasy that doing this
would create pressure on the Architecture group to provide an
official version, but I think that the clear value of making an
unofficial version available to our customers (with appropriate
disclaimers) would be well worth some unease.

Back in 2013, I also created a related document, the "HLASM
Instruction Set Overview", which provides summary tables of all
instructions (again as of zEC12) grouped by category showing
which combinations of operands and functions are supported, using
different colours to show which instructions were available for
each machine type. There is a somewhat better chance that I might
be able to make that available externally in some way, especially
as I am now the HLASM team leader (following the retirement of
Sharuff Morsa back in June).

Remember that it's always much easier for us to make things
happen for customers if customers are on record as actually
asking for them.

Jonathan Scott
HLASM team
IBM Hursley, UK

Reply via email to