W3C home > Mailing lists > Public > www-xsl-fo@w3.org > March 2002

RE: Pushing the Limits

From: Arved Sandstrom <asandstrom@accesswave.ca>
Date: Sat, 23 Mar 2002 10:23:43 -0500 (EST)
To: "David Tolpin" <dvd@renderx.com>
Cc: <www-xsl-fo@w3.org>
Message-ID: <NIECINNOJOOPPAIFLMIGCEGGCEAA.asandstrom@accesswave.ca>
That's a really good question, David, and I'm not surprised that you asked
it. :-)

The reason I'm worrying at these niggling issues is because that is where I
currently am with my other formatter project (MiniPlug: xslfo-proc on
SourceForge). More precisely, this is where I am with the Perl prototype for
that project. You are already aware that I am doing this. I am taking a hard
look at my manager design, and right now specifically I am dealing with
keeps. This is where this example came up.

I had already decided to do what you guys do, in respect to look-forward. I
didn't know you had such an internally configurable parameter but I can't
say I am surprised. Yes, absolutely, you make a very good point - what if
the keep-together, for example, was badly thought out, or the user didn't
realise that the flow object is too big for _any_ page master?

I am currently waffling but I think I will settle on 1 or 2 pages of
look-ahead. No more. My page-sequence manager class (Perl module, really,
hate to call it a class, per se) goes (or will go) into a different mode
when it sees keeps, and it starts to buffer layout, and if after a few pages
there is no end in sight to things like empty page production, it'll either
call a halt to the entire affair or do something like what XEP does.

You have raised what I am afraid is the real complication here: did the
circumstance arise because it was intentional, or unintentional?

In a nutshell, here's what I would do in this situation (that is, that is
how I intend my formatter to work): my look-forward setting would permit
production of the single blank page, and a warning would be produced. If
more than one blank-page were to be produced by a slightly different
page-sequence-master than the one I have in the example, I would terminate
processing, again with a warning.

I don't think the XEP solution is unreasonable - the odds are probably that
9 times out of 10 this happened because of error, not because of some XSL
wizard with significant formatting-fu - but I'd like to see a warning issued
if you exercise this fallback. I don't recall seeing one, although a careful
perusal of the XEP console output does indicate some page-level layout
decisions related to this situation.

And maybe formatters could have at least 2 modes, configurable also -
"friendly" and "strict".

Regards,
Arved

-----Original Message-----
From: www-xsl-fo-request@w3.org [mailto:www-xsl-fo-request@w3.org]On
Behalf Of David Tolpin
Sent: March 23, 2002 10:55 AM
To: Arved Sandstrom
Cc: www-xsl-fo@w3.org
Subject: Re: Pushing the Limits


Arved,

How many pages should the formatter look up forward before it decides that
there is a trouble in the flow, not a cleverness in the layout-master-set?

This parameter is internally configurable in XEP. If I have to skip a page
because it is empty, is it empty because the stylesheet developer wanted
it to be such by specifying small a region area, or is it happened
unintentionally
because a flow object was too big?

If I have to skip two pages? What if I have to skip 100 pages? How many?

David
Received on Sunday, 24 March 2002 07:49:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 3 October 2007 16:06:09 GMT