Re: Guile Startup Time Suggestion

[Andrew Archibald <aarchiba@calum.csclub.uwaterloo.ca>]

hjstein@bfr.co.il (Harvey J. Stein) writes:

> However, it's much easier to write specific dumpers for specific
> situations than to write a generic unexec.  In your particular case,
> why not write something to do a binary dump of the DFA to a file, &
> read it back in when you restart?  Pickling & unpickling specific
> data structures when you know their pointers is *much* easier than
> writing a generic unexec.

You're right.  This does make things a bit limiting as far as
extensions go (that is, using guile extensions as opaque data) --- if
a data structure is to be pickled, then it must be possible to pickle
anything it refers to.  This almost has to be a linguistic thing.  If
"read" could be extended to handle this for user smobs, this might
almost be feasible. 

> I don't understand a couple of comments you made, though:

I should point out that I don't really understand the lang code;  I
just tuned it up to work on a modern guile. 

> 1. Are you saying that most of the time is spent in a C fcn which
>    initializes the DFA as opposed to Scheme code calling C code
>    over & over to generate the DFA?

There is at least one long interval in which we're deep inside C code,
grinding away. 

> 2. I wrote a Pascal header parser using Bigloo and I never had
>    anything even remotely resembling 15 minutes to startup.  And this
>    was before compilation.  Why is the language parsing code taking so
>    long to generate the DFA?

I don't know how elaborate your header parser was.  This is the parser
for CTAX, a language syntactically about as complex as C.  The lexer
is part of the startup time, and the laily-constructed parser data
structures are the other part.  It's worth pointing out that the
machine is an ancient 386 with 8 megs of RAM, so it's not the
speediest, especially when swapping. 

You can test this yourself with the guile-lang-allover package.  It
will boot up quickly but the first expression it tries to ctax-read
will take some computing. 

I should look at Bigloo; maybe the lang package should be rewritten.

Andrew