T.R | Title | User | Personal Name | Date | Lines |
---|
859.1 | | VNASWS::GEROLD | You're my favourite waste of time ! | Mon Aug 31 1987 07:56 | 7 |
| I got bitten by that, too. The source of all evil was in the SYSUAF.DAT,
where the directory was specified as <dir>.
So as a quick workaround change it there.
Actually I think there should be things higher on the priority list.
Gerold
|
859.2 | Glad to hear about V1.1 enhancement | STAR::GUISSO | Ridendo dicere severum | Tue Sep 01 1987 04:36 | 10 |
| I'm glad to hear that V1.1 will address the problem. My files have
many angle-bracketed directory specs. Any coding "workaround" that
requires the writer to define a complex macro is really unacceptable.
In the meantime, I'll just use angle brackets for directory specs,
as my tech reviewers require, and I'll notify my editor to expect
appropriate "undefined tag" error messages in my bookbuild log files.
I refuse to pollute my source files with spurious tags, because
I don't want to confuse the next writer/maintainer.
Lazarus
|
859.3 | when is a <rose> not a <rose> tag? | VAXUUM::KOHLBRENNER | | Thu Sep 03 1987 13:03 | 33 |
| The improvement/fix that will come for v1.1 will address the
problem of being able to supply file specifications that use
angle brackets for the directory part of the file specification.
That is what I defined as problem 1. Thus, you will be able
to specify a file spec on the DOCUMENT command, or in the
argument to an <INCLUDE> tag that has an angle-bracketed
directory name.
There is no fix/improvement coming for a file specification that
you supply as *text* in your SDML file. The SDML language will
continue to use < and > to surround tagnames. Thus any angle-
bracketed sequence of characters that meets the rules for a name
gets recognized as a tagname and gets diagnosed as undefined if
it is not a tag. There is simply no way for DOCUMENT to determine
that the <TABLE> 'tag' in the following example is not a tag:
<code_-example>
NODEXX::DEVICE:<TABLE>FILENAME.TYP;5
<endcode_example>
We *may* define an easier way to write <literal>(<), so that you
do not have to define your own "complex tag" to do it, but you
will have to do something to avoid having the tag translator see
<table> as a tag.
There is a limit of 30 warning messages after which the tag translator
notifies DOCUMENT not to proceed to the next step in processing.
So letting all those directory names that look like tags go through
with warning messages may prevent processing past the tag translation
step.
bill
|
859.4 | How about <<TAG>> | EAYV05::WESTON | Telecomms, the key to Tomorrow | Fri Sep 04 1987 12:16 | 11 |
| As a suggestion, how about looking at the use of double angle brackets to
tell the tag translator that the element is not a tag and the content,
including the one pair of brackets is a literal?
For example, <<TEXT>> is equivalent to <literal><TEXT><endliteral>
This has the logic of treating "<" as an escape character which is negated
by the immediately following "<" and vice versa at the termination. This is
similar to the DLE character in comms protocols
John
|