Re: On a programming manual for Lout

[Jeff Kingston]

I have no objections to someone writing a programming manual for Lout
as long as it is not a panegyric.  Here are three areas where I think
the semantics of Lout, or at least its implementation, are problematic:

* Tables - I don't understand how spanning columns work together
  with all the other Lout features, and I'm the author.

* Macros - these look like ordinary symbols to the end user, but behave
  much worse because they don't combine cleanly.

* Scoping - another area where the semantics are obscure, at least to
  me, to the point where there is still a scoping bug in the database
  reading code, fortunately rarely encountered.

You could add galleys to this list, but at least there the complexity
is more or less justified.  Think of troff's diversions and traps,
for example, and galleys don't seem so bad.

In addition to things in the language that are bad, there are things
not in the language that should be there.  Three examples:

* It is not possible to do things which are taken for granted in
  programming languages, such as arithmetic, string functions, etc.
  An irritating consequence of this is that reference lists in Lout
  can't adapt to all the (admittedly absurdly overspecified) styles
  required by publishers for reference lists.

* It is not possible for the Lout programmer to enhance the "style
  information" (the equivalent in Lout of Unix environment variables).
  This has led to some pretty disgusting stuff in packages (equations and
  syntax diagrams spring to mind) where such enhancement is needed.

* Lout is not object-oriented and this leads to a ridiculous
  amount of code duplication.  All the code for figures is copied
  textually and names changed to create the code for tables - horrible.

I must say I feel a bit worried about the idea of people going away and
writing a lot of software on such a flawed base as Lout is.  I know that
most document formatting systems are a lot worse, but many modern
programming languages are a heck of a lot better.

It's a curious state of affairs that people should be still discovering
Lout and rating it as the next cool thing, when the language itself is
already showing its age (16 years).

Jeff Kingston