> 1. Emacs considers <p> and <list> within <argument> as invalid ...
Can you provide some more information? Flavor of emacs, version of
Emacs; flavor of sgml mode; version of sgml mode software (e.g.,
psgml)? What do you mean "emacs considers invalid" -- emacs does not
allow insertion, it finds it as an error when you enter C-c o, or it
is an error on validation (C-c v)?
> 2. Notation of empty elements ... With XML TEI, ... <pb> or <pb/>?
Either "<pb></pb>" or (preferably, imho) "<pb/>" should do, I think.
> Should I completely switch over to <pb/>?
Depends on the details of your workflow and project, but my first
reaction is "I would" (although in fact, I haven't -- but I think
for most projects XML is probably more appropriate).
> What can I do if I start off the transcription with a word processor
<soapBox>Don't. Word processors do not have a role in the creation of
plain-text files. Dedicated editors (e.g. ArborText, Author/Editor,
XMLSpy, etc.) are OK, of course.</soapBox>
> (I'm not happy using emacs for this)
Makes me think you need to learn more about emacs :-)
> ... already have <pb n="175">, which it is impossible to just
> search and replace.
Ah, but it is not only possible, it's easy to search and replace.
Even without specialized SGML->XML software (which I've written for
the WWP, and thus should be something we could modify for use at any
particular TEI project pretty easily), many of the changes needed are
quite easy; the trick is to use regular expression search and
replace. In particular inserting slashes into PB tags:
M-x query-replace-regexp RET; usually bound to C-M-%, but not always
<pb[^>]+ RET ; at "Query replace regexp:" prompt
\&/ RET ; replacement at "with:" prompt
;; Note: query (emacs' word for "search") finds all strings
;; that start with "<pb" (case insensitive depending on your
;; setting of case-fold-search) up until, but not including,
;; a ">". This includes strings inside comments, attibute
;; values, and ignored marked sections, not mention any
;; "<PB-and-J-sandwich" tags you might have. As always, be >
;; careful using non-\(SG\|X\)ML aware tools on your data.
;; The replacement pattern "\&" stands for "whatever string
;; I (emacs) matched".
;; As far as I can tell, C-M-% was not bound to query-replace-
;; regexp in version 19, but is in version 20 it is. At the WWP
;; we use it so often we have bound it to M-#, which is easier
;; to type.
To learn the syntax of regular expressions read the doc for your
particular editor. (Or, better yet, if you're going to Extreme Markup
Languages in Montreal in a few weeks (note that the link off of
Mulberry's page is now broken; the new one is
http://www.extrememarkup.com/extreme/), sign up for my emacs & XML
class!) For me that's chapter 12 of the GNU Emacs Manual, 13th
edition (manuals are availabe in print from the FSF for quite
reasonable prices -- see the bottom of page
http://www.gnu.org/home.html for contact information; online from
that same page; or better yet, because you're sure the manual is the
right version for your emacs, by entering
C-h i ; enter emacs' internal documentation system
m ; request a particular info menu item ...
emacs ; the node we are interested in
RET ; brings up the emacs manual!
C-s ; search
search ; for the word "search"
RET ; stop searching for "search"
RET ; go to the chapter we just found
> [Note to the editors of the Guidelines: I couldn't find anything on
> this in the Guidelines (all the versions I've got). A paragraph on
> notation - forward slash or no - would be extremely helpful.]
Caveat -- I am not one of the editors. On the other hand, I do know
that the P3 Guidelines (1994-08) were published long before XML
(1996-11) existed. I believe future versions will be more XML-
> 3. Modification of an element's attribute with the Pizza Chef
> ... I successfully managed to add an attribute ... using the Pizza
> Chef. With one tiny restriction ... I wonder what went wrong.
(To the tune of "Rainbow Tour" by Andrew Lloyd Webber and Tim Rice:)
Let's hear it for the Pizza Chef
It's been an incredible success
We weren't quite sure, we had a few doubts
Would it be easy enough?
And the answer is--
Can you provide the details of what you did to get the two declarations
of one element result? (Although you may want to wait a bit; others who
know more about the Pizza Chef's working may be able to answer you with
just this information.)
> 4. Number of displayed errors in emacs
> Emacs restricts the number of errors to 200 ... It says this can be
> changed with the -E option [?]. I can't find anything about such an
This sounds very much like you are using nsgmls as your external
parser, and it (not emacs) is what is complaining about the number of
errors. -E is indeed the switch for changing the maximum allowable
errors that nsgmls will process. So I'm betting you enter
and emacs puts something like
nsgmls -s -wmin-tag my.file.tei
into the minibuffer after the prompt "Validate command: ". That is
the command string that emacs will send to the shell to execute; it
will take the output from the shell and stick it into the "*sgml
validation*" buffer for you to check out. The good news is that you
can modify that command before sending it off to the shell. Just back
up to where the switches go, and enter " -E512 ", then press RET, and
be prepared for a *lot* of errors. BTW, to ignore a whole bunch of
errors that I don't want to deal with in the *sgml validation* buffer
(although the empty-tag errors are so easy to fix, see above), I use
the flush-lines command. E.g., if I had millions of errors that
nsgmls:test.xml:7:23:E: end tag for "i-am-empty" omitted, but
OMITTAG NO was specified
nsgmls:test.xml:7:12: start tag was here
I could issue
C-x 0 ; make *sgml validation* the current buffer
M-x flush-lines RET ; a wonderful cmd; opposite is keep-lines
[regexp] RET ; at "Flush lines (containing match for
The regexp itself won't fit on one line, and a variety of regular
expressions could be used. So substitute something like
end tag for .* omitted, but OMITTAG NO was specified
.*start tag was here
(which I just tested) for "[regexp]" above. In many flavors and
versions of emacs you won't get to see the begining of the regular
expression string after you enter the newline (C-q C-j) -- don't
panic, just keep going.
> 5. Declaring character entities with an ISO set
> ... I've tried:
> - the 'standard invocation'
> <!ENTITY % ISOlat1 PUBLIC "ISO 8879-1986//ENTITIES Added Latin 1//EN">
> Emacs complained about the bit following the ">". I deleted it. It
> didn't work.
Although I don't have enough information to get this to work for you
(we need to know where your ISO files are stored, are they SGML or
XML, where your catalog file is, etc.), I can give you an overview
that might be helpful.
The stuff between "<!" and ">" (MDO and MDC in SGML abstract syntax
terms) tells the system "I am declaring a parameter entity named
'ISOlat1'; it's value is whatever you find the formal public
identifier "ISO ... //EN" means". Typically the system (at least, in
the case of both emacs/psgml and nsgmls) resolves the FPI by looking
in a "catalog" file for a string that matches the FPI, and grabbing
the file it is associated with. The catalog entry might typically
"ISO 8879:1986//ENTITIES Added Latin 1//EN"
The "bit following the '>'" appears to be in this case some
whitespace and then "%ISOlat1;", which is a reference to the
(recently declared) parameter entity reference. It says to the system
"replace me (the reference) with whatever has been declared as my
value". So many systems do not actually try to get that which is
declared as the value of the parameter entity until it is actually
referenced. I.e., when the software gets to the
<!ENTITY % name PUBLIC "fpi">
line it will check for obvious mistakes (like illegal characters in
'name'), and try to find the path to the file of interest (e.g.
"/usr/local/.../ISOlat1"), but won't actually try to open the file.
When it gets to the reference ("%ISOlat1;"), it then tries to
actually open the file. If the file's not there (or unreadable, or
has characters out of your declared encoding, or whatever), you get
an error flagged at that point (i.e., flagged at the reference, not
> - I replaced the Public ID with a System URI
> <!ENTITY % ISOlat1 SYSTEM "a:/isolat1.ent">
> Didn't work either.
Did you put the reference ("%ISOlat1;") back? If so (and if
"a:/isolat1.ent" is actually the right file :-) seems it should have
> - Then I just copied one of the entity declarations from the ISOlat1
> file (and deleted the comment after the replacement characters):
> <!ENTITY auml SDATA "[auml]">
> Emacs gave a warning about "SDATA", which I deleted.
XML does not know about SDATA (an SGML construct). I have a complete
set of ISO entity sets that have been "XMLified" and which use ISO
10646 (with thanks to Rick Jelliffe, who made the files -- all I did
was fix a few bugs and combine them into one big file). I'm not 100%
sure that doing so conforms to the copyright restrictions ISO places
on the files, though; thus although I'd like to make them available,
I'm not sure I should. (Nor do I know where I'd put them, but that's
another question.) Anyone from the ISO legal department reading this