T.R | Title | User | Personal Name | Date | Lines |
---|
600.1 | references get resolved before page layout | VAXUUM::KOHLBRENNER | | Tue Jul 07 1987 08:26 | 13 |
| Unfortunately, there is no way to refer to page numbers.
Currently all symbol definition and all resolution of symbol
references takes place in the tag translator, before any
page layout has taken place.
The references to page numbers that do occur (in the table of
contents and in the index) take place during page layout of the
body of the document. However, those references get formatted
on separate pages. Once TeX has formatted a page, it never has
to return to that page in order to insert a page number.
I don't know if TeX can do such a "fix up" operation.
|
600.2 | Let TeX do the work | CUPOLA::HAKKARAINEN | This too shall pass | Tue Jul 07 1987 09:06 | 11 |
| If Document could get TeX to write out a symbol table, containing
reference markers and the appropriate page numbers, then TeX could
re-read that table on a second pass. Accurate page-based
cross-references will require at least two passes through the text
formatter. Those additional passes need not occur during the same
session. DSRplus, for example, checks to see if its symbol table
(.brn file) is older than the main (.rno) file; if so, it updates
the symbol table.
This feature would also permit numbering such as ``Page 1 of 5,
Page 2 of 5'' and so on.
|
600.3 | coding for disaster | VAXUUM::KOHLBRENNER | | Tue Jul 07 1987 10:23 | 18 |
| It seems to me that it is still a little messy. In order to make
up the page the first time around, TeX has to estimate the space
that will be taken by the reference to the page number. It
doesn't know if the page number is 5 or 555. If there are
many such references on a page, then the amount of "stretch"
that will be required could be quite large.
Then on the second time around, TeX will be replacing references
to symbols with the actual page numbers and it has to keep all
the page breaks at exactly the same place that they were the
first time through. THus it has to squeeze or stretch its initial
layout of each page to accomodate the new material on that page.
One could easily build a pathological case that would produce a
page with a lot of white space on it or a page with everything
crammed together. Granted, most real-life situations would not
cause a problem. But the algorithm has to handle ALL situations,
not just the simple ones. I think the code to handle the patho-
logical situation might dwarf the rest of the code.
|
600.4 | LaTeX can do it, TeX does it, why not Document? | IJSAPL::KLERK | Theo de Klerk | Wed Jul 08 1987 04:29 | 15 |
| Re .1:
Yes, TeX can do this, in fact LaTeX does it and that's why I started
to use this feature and appreciate it.
Every \label{symbol} creates an entry in the .AUX file, that is read
on the second pass through the source. Admitted, this means processing
the document twice. On second pass round it uses the .AUX file to
resolve any \ref{symbol} or \pageref{symbol}.
The table of contents entries are written in a .TOC file and processed
the same way on second pass. Idem for figures and tables. This functionally
is the same as with Document.
Theo
|
600.5 | wishlist | CLOSET::ANKLAM | | Wed Jul 08 1987 12:18 | 4 |
|
This came up quite a long time ago in the old DOCUMENT NOTES file,
and it is on the list to look at for a future release.
|
600.6 | Bounded iterations | BUNSUP::LITTLE | Todd Little NJCD SWS 323-4475 | Thu Jul 09 1987 00:53 | 15 |
| Just as an aside for future consideration, an approach that is
almost guaranteed to work in ALL cases but work slowly in the pathalogical
cases is to recalculate the symbol values on the second pass. If
any one of them changes due to TeX making new page boundries, you
run yet another pass. I say almost because I suppose its possible
to get into a pathalogical case that causes oscillation, but I'm
not sure you could create such a case, and you could punt if you
didn't get convergence within say 3-4 passes. In most cases, you'd
get convergence on the second pass and only in the most bizarre
cases would it take more than 2 (like 100's of references on a single
page to a page number that went from N to N+1 digits, and even then
thats only adding 100's of characters to the current page which
still might fit.)
-tl
|
600.7 | let's try that again... | VAXUUM::KOHLBRENNER | | Thu Jul 09 1987 09:23 | 7 |
| re: .6
Maybe we could get TeX to emit beeps and type @#$%^&%$^ words
on the terminal as it began the 2nd, 3rd and 4th iterations...
bill
|
600.8 | You're very welcome | BUNSUP::LITTLE | Todd Little NJCD SWS 323-4475 | Fri Jul 10 1987 01:22 | 15 |
| re: .7
I can only tell you that as a user of the documentation DIGITAL
produces, I find the lack of page number references distinctly
annoying. Nothing like having to thumb around through the pages to
find what you've been asked to look at, or turn back to the TOC and
scan for what you were looking for just to get the page number.
If you have a better suggestion or method that will generate page
references properly, go right ahead, I was only trying to suggest a
method that has been used before for similar problems. For that
matter, does anyone know offhand how LaTeX does it and whether it
handles the bizarre cases.
-tl
|
600.9 | how will the wizard feel on the Nth pass? | VAXUUM::KOHLBRENNER | | Fri Jul 10 1987 10:37 | 22 |
| I think what TeX will have to do is exactly what you specified in
reply .6. And probably 99% of the cases would not result in
additional passes beyond the second pass.
I think I may be over identifying with the executing code in TeX.
I was imagining that code
going through the second pass and discovering that another
pass would be needed, gnashing its teeth, looking at the clock,
sensing the drumming of the user's fingers on the edge of the
keyboard, while s/he waited for output, and beginning to mutter
about how hard it is to satisfy these users... Then when the
next pass came around and yet another pass was needed...
With all those wacko error messages that TeX used to put out,
you have to admit it does seem to have a personality... I can
imagine it getting a bit miffed at having to repeat itself just
for a simple page reference. Most "wizards" that I have met
do not seem to have a lot of patience with repetitive tasks.
:-)
|
600.10 | So father so son | IJSAPL::KLERK | Theo de Klerk | Fri Jul 10 1987 11:28 | 5 |
| > wacko error messages that TeX used to put out,
What do you mean? Doesn't the child talk like his father?
Theo
|
600.11 | DECspeak, not TeXspeak | VAXUUM::KOHLBRENNER | | Fri Jul 10 1987 12:43 | 3 |
| Well, David has been working very hard to have the child
merely mumble the parent's comments (inaudibly).
In the meantime, the child has been taking a crash course in DECspeak...
|
600.12 | Sell it as a "feature". | BUNSUP::LITTLE | Todd Little, New York Area SWS, 323-4475 | Mon Jul 13 1987 11:51 | 8 |
| When this ever gets done in DOCUMENT, you can make a note in the
release notes about the sensible use of page references and mention
that DOCUMENT in an attempt to prevent bad writing goes out of its way
to punish you for excessive page references!
-tl
PS I've got a bottle of valium earmarked for the wizard.
|
600.13 | A different side of the problem. | VAXUUM::CORMAN | | Tue Aug 11 1987 15:54 | 11 |
| In the meantime, the writers in the Layered Products group have
been told that, not only should we not reference the page number
of information in VMS manuals, but we should not reference the
specific VMS *book* itself. This is NOT because VAX DOC doesn't
yet offer the page referencing feature... it's because the
VMS docset is so big and transient that the referenced book might
not be there tomorrow.
Of course, that's a discussion for a different notesfile...
Barbara
|
600.14 | So what? | IJSAPL::KLERK | Theo de Klerk | Tue Aug 25 1987 10:33 | 21 |
| Simply because VMS does such a lousy job in sticking to titles,
divisions etc. it does not mean other documents, more stable than
the VMS set, should not be able to reference to pages.
I want to be able to do that, even if VMS doc doesn't.
Theo
Footnote:
Having been in the field for quite a while + stood in front of classes
for 4 years I know customers find VMS docs increasingly difficult to
find all the information on a certain topic. It's getting spread all over
the place with a master index that is often wrong. Perhaps as well we
don't mention particular page numbers.
"Any resemblance to existing pages is purely coincidential since they
do not yet exist"
(paraphrased end credit title to "Thunderbirds are Go!" feature film)
|