Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!pasteur!ucbvax!husc6!lloyd!kent
From: kent@lloyd.camex.uucp (Kent Borg)
Newsgroups: comp.sys.mac
Subject: LightSpeed Pascal pretty-printing
Summary: It's a feature--really.
Keywords: parse token pretty printing Lightspeed Pascal LSP incremental compiler
Message-ID: <244@lloyd.camex.uucp>
Date: 3 Nov 88 15:41:46 GMT
References: <1151@usfvax2.EDU>
Reply-To: kent@lloyd.UUCP (Kent Borg)
Organization: Camex, Inc., Boston, Mass USA
Lines: 54

In article <1151@usfvax2.EDU> pollock@usfvax2.UUCP (Wayne Pollock) writes:
>Below is a LSP program (re-pretty-printed to save screen space; why oh why
>can't I turn this option off?).  ...

Because it is not an `option'.

Rich explained this a while back.  What I remember: One of the reasons
LSP has such a fast turn around is that when you say `Compile' a lot
of (all of?) the parsing of your program is already done, it happened
when you typed the `;' for that line.  The compiler is storing your
program in some tight, tokenized, internal form.  This has the natural
side effect striping out formatting information that the compiler has
no use for.  Remember back when you were playing around with
interpreted BASIC?  You typed in `print' (or `?') but the listing said
`PRINT'.  Similar effect.

Certainly the pretty printing could be avoided without loosing the
fast turn around and source code debugging LSP does so very well, but
I can see how these changes would be far from trivial and not without
cost in time and space.  That coupled with idea that pretty-printing
might be a _feature_ helps explain why it is the way it is.

LSP 2.0 is supposed to allow the user lots of opportunity to twiddle
the pretty printing parameters--but it is still part of the product.

Personally, though I like the pretty printing, I do have some gripes:

It will sometimes `kidnap' an end statement from old code when I am
adding new code, leaving me unbalanced (more than usual I mean).  In
this case it is looking at my work too soon.  

On the other side it is sometimes hard to get it to reconsider a line
once I have fixed a typo.  This was a particular problem for a class I
recently taught at the Boston Computer Society where LSP was used for
examples.  Students would have a perfectly fine Pascal statement
sitting on the screen, but the editor was still upset over an earlier
typo.  In this case it is looking at our work too late.  

This incremental stuff is complicated, and when I think about it, I am
impressed with how gracefully they do it.

Further consideration for all you grumblers: LSP is great and there
are ways they could make it still better.  Getting rid of the pretty
printing will not be trivial.  Do you really want them to take time
away from making substantial improvements just to fix this?  Wouldn't
you rather get inline assembly code, or something else that actually
adds power to the product?  If so, say so, and don't complain so much.

(Personal LSP request: Editor support of arrow keys, option arrow
keys, command arrow keys, shift option arrow keys, etc.)

Kent Borg 
kent@lloyd.uucp 
or 
hscfvax!lloyd!kent