The LAC HyperGuide
Division meetings/meet1es, 10.9.1995
Meeting 1:
3-6 September, 1995 in Mauritzberg, Sweden
Notes by Erik Sandewall re discussions and decisions at the meeting
as a whole.
The purpose of the meeting was to (1) discuss the possibility and
appropriateness of developing a Hypermanual for Logics of Actions
and Change within the present Esprit working group
Logic and Change, and (2) to set the guidelines for the
continued work. The material which had been distributed for the meeting
was (a) a partially developed WWW structure suggesting how the
Hypermanual could be organized, and (b) a list of concrete questions
to discuss, mostly of a practical character.
The following is a list of concrete decisions at the meeting,
organized in a somewhat logical order (not equal to the order in
which the decisions were taken). The present version of the list is
tentative, pending approval from those present. The numbering is
for ease of reference. Questions marked in italic indicate points
where I am uncertain about the meeting's decision, or the decision
was ambiguous or incomplete. Feedback is welcomed on those points.
If no feedback is received, I'll leave it as it stands.
- The participants wanted to go ahead with the project, but
restricted to the "Hypermanual" as such. The related proposals
for a depository or archive or technical reports, and for a
bibliography containing commentary information, were encouraged
but not included in the scope of interests of the present group.
- The name of the structure shall be the hyperguide for
logics of actions and change, to be abbreviated as
the LAC HyperGuide.
Capital G in HyperGuide appropriate?
Thus the term "Hypermanual" is abandoned.
- The importance of making full use of the new medium's
possibilities was emphasized. The "table of contents"
of the hyperguide (as distributed for the meeting) should be
understood exactly as that - a map where one can see what is
included - but it is foreseen that the actual order of reading
the contents will be different. Some parts of the structure
should be seen as a partially ordered graph, which can be
explored by the reader with or without the help of "guide"
information. The order in that graph indicates definitional order
(definitions of a term precede its use), but the reading order
may often be in the opposite direction, namely if the reader has
a general knowledge but then later needs to go back to
definitions for some details.
- The structure of the hyperguide's essential contents were
discussed in depth. It resulted in the decision to use the
structure shown in
[meet1-contents]
Note: the names of logical subdirectories which were decided
on the afternoon of the last day have not yet been
incorporated in this file. Also, some of the headings are missing,
for example for "Planning". Will be corrected.
- Erik Sandewall and Antonio Porto will together develop the
material on "Frameworks". They will also produce an initial
scetch for the material on the first three subdivisions of
"Logics".
- Jim Cunningham and Camilla Schwind will together develop
the material on "Logics".
- Jacqueline Vauzeilles and Marcel Masseron will develop material
on linear logic (to be reported to Jim and Camilla) and on
planning.
- Camilla Schwind will also develop the material on "Update".
Marco Schaerf indicated an interest in the same topic, but had
difficulties finding the time. Shall I include you here, Marco?
- A standardization of the notation for mathematical and
logical formulae in the hyperguide was discussed, based on a
table of symbols used in the book "Features and Fluents".
The meeting decided to use the set of symbols specified in
[notation]
Please check again that I got it right.
- It was agreed to introduce a reference language, that is, a
formal language with a precise syntax, for expressing
scenario descriptions involving actions and change in a
uniform and high-level fashion. The name of that language was
left open.
- How about calling it the HyperGuide Reference
Language (HGRL) ? I provisorically use that term in
what follows.
- The HGRL will be used as the primary tool for expressing
scenario examples, and the various specific ("integrated")
approaches which are found in the contemporary literature
will be defined and analyzed in its terms. An operational
semantics will be defined for the HGRL, using if possible the
methods in "Features and Fluents" with the adaptations that
may be found necessary.
- The HGRL will be seen as an alternative to the script-A
family kind of languages, and will be characterized by a
closer relation to logic. Its syntax overlaps with first-order
logic, but involves some constraints and some extensions with
special operators. The HGRL will be developed by Erik Sandewall
and Antonion Porto, starting from the guidelines that were
formulated at the meeting. These are being written down in
[frameworks/hgrl].
- The need for a collection of representative examples was discussed,
with Jim Cunningham as the chairperson. The notes from this discussion
are in a separate file,
[meetings/meet1jc].
- Hyperguide segments with a substantial content shall be
constructed by analogy with ordinary articles, that is, they
shall have an author (set of authors) and a title. For
reference purposes, they can then be treated as analogous to
articles in a "collection" type book.
- The hyperguide as a whole, and larger parts of the hyperguide,
can then be referenced as a composite work with one or a few
editors.
- The permanent availability of earlier generations of
hyperguide segments was discussed. It was agreed that such
permanence is very desirable, but that there are a number of
practical difficulties. (It was not agreed exactly how to solve
them). Referencable hyperguide segments shall be marked with a
version number and a month of issue; either
one of these information shall be sufficient for correctly
identifying and retrieving the segment version in question
from the archive.
- The importance of a good introduction to the topic
of the hyperguide was emphasized.
- (I proceed now to the list of fairly practical decisions).
- It is desirable to have a virtual host name for the hyperguide as
a whole. The Linköping participants will investigate how this
can be done. Not yet completed, work in progress.
- We need an addressing scheme whereby addresses to pages within
the hyperguide can be kept unchanged as parts of the hyperguide are
moved from one host to another. It was agreed to use a syntax as
in the following example for references within the hyperguide:
/hg-lac/logics/modal/
where the first part (hg-lac) shall invoke a script in
the current host which maps the rest of the path to a real one,
and then returns the intended file (if it is in the same computer
system) or requests it to be sent to the user (if it is in another
computer system). The work of implementing this was considered
small. Lars Karlsson promised to try to do the job until the end
of the month of September.
- The identifiers in access paths shall be chosen as real English
words; "mnemonic" identifiers are to be avoided.
- Two formats are to be used for hyperguide segments ("files"),
namely HTML and postscript.
- Postscript pages are to be generated from Latex.
- The Latex2e variant of Latex is to be used.
However, since then I have used it, and it is hopelessly slow
for short postscript files. It seems to do a lot of initialization;
maybe the overhead is less damaging for longer files, but since
we want to emphasize short files I think we have a problem here.
I would propose to change the decision to say that we shall
try to write segment source and macros so that they can very easily
be moved between older Latex and Latex2e. This will allow using fast
Latex while writing.
- The heading of hyperguide segments ("files") in printout was
decided. The file
[frameworks/hgrl].
shows an example of the decided format.
Except I did not manage to get rid of the page number on
the first page, which was intended. Anyone knows how to do that?
- We discussed the fact that postscript fonts that display well on
the screen come out too thick on paper, and vice versa. It was
agreed to handle this by keeping postscript files in two versions,
namely, a .ps version for printout and a .dvi
version for display. Some concern was however expressed regarding
the lacking standardization of .dvi.
After the meeting I have observed that the ECCC (colloquium
on computational complexity) originally used the postscript + dvi
scheme, but that they discontinued the dvi:s after a year.
The following items are not yet completely documented in this version
of the minutes. They will be included as soon as possible:
- The agreements about the syntax and font choice in the HGRL
have not yet been written out in full.