DynaText

DynaText is a hypertext system based on SGML and a large number of Graphics standards. It comes from Electronic Book Technology

Basically portable, posix/X conforming.
Uses SGML format for document storage
Large variety of graphics formats
Full text indexing included.

Dynatext is now available for Suns - (os4.1 or higher, 10MB nnedeed for full installation), on Sun-2 3.5"floppies or cartdidge. Production version 1 now available (March 91). Mr. Schmidt is getting us one on cartirdge (13 March 1991). Evaluation price for 45 days: SFr750 (CERN half price). Under evaluation.

Aparently Windows-3 version is avilable beta test only - Chris jones has one.

First impressions

We (Giuseppe Mornacchi and I) installed the product on a sun4 running openlook, but noticed that the motif version supplied with it works fine (even under openlook).

The first disappointment with dynatext was that it is not an authoring system. It requires SGML to be prepared in advance. It then builds a book with it, which involves building indexes etc. After this process, the book can be browsed.

A strange feature is that the licencing is done on builds: each time you rebuild a book, your license is decremented! Therefore, it is not really suitable for fast-changing cooperative work: it is targetted at centralised publication.

The browsing is basically good, very similar to DEC's bookreader product. One can have more than one window on hte same document, but by default following a link replaces the text in the window one has.

Searching

I found searching difficult at first, because one needs access to three windows at a time. The table of contents window starts the search, a lookup window has the parameters, and the results are highlighted in the text window. As under openlook I found I was always losing windows on the screen, it would take me some time to find the results of my search. The default positions for the table of contents and lookup windows seem to be on top of each other, which didn't help. Windows do not pop to the surface when something interesting happens in them.

The search result is neat, however, in that the contents page is annotated with the number of occurences of the pattern in each chapter, section, etc. (The contents page is an outline viewer which allows contraction and expansion of the nesting).

Links

Links are not represented by highlighted text, but by icons in the margin (in the case of the documents we read, the right hand margin, but this is defined in a style sheet.) This involved the use of macros (SGML entities) for phrases like "see table linked at right" for the browsable version, but "see table below" for the printed version. This meant that the flow of the text was broken by such references, so in practise the document only contained a few links. In their example documentation, this was infuriating: there would be a reference to another part of the document in the text, but no link to it!

There are links within and between documents. One can, in the DTD, specify the effect of a link using a number of predefined "scripts". I don't know how flexible this is and whether it would allow us to add WWW-style networking capability.

General

As authoring is not supported, neither is annotation, though I understand from EBT that a future version will support annotation (only - not full authoring).

There was a steady chatter to the standard error output of little error messages, which didn't seem to affect anything visible in the windows.

Out attempts to get a wider test public by displaying dynatext on a Vaxstation running the (field test) X11R4/Motif failed: the font "fixed" could not be found. We put the blame for this on DEC rather than EBT.

Interesting possibilities for Cern would hinge on someone producing the DTDs and style sheets for CERN SGML files and/or WWW HTML files. Perhaps the LHC test beam team will have time to try this.