cl-info revision



Robert Dodier writes:
> I am working on a revision of the cl-info code (which powers
> the describe function). The primary goal to make describe run
> faster; it is too slow on slow processors (e.g. mine).
> A secondary goal is arrange the searchable topics so that
> different kinds of searches can be implemented
> (e.g. exact match vs inexact, title vs category vs keyword).
> 
> The major part of it is to search for topics in a list of topic
> strings, instead of carrying out searches on the info file itself.
> The topic strings are keys for a hash table in which the
> associated value tells what info file to read, the byte offset
> of the item, and the length in bytes. Then retrieval is very fast.
> 

Very good news!  Do you plan to include this into 5.10 ?

> The hash table is static; it is built once, and then loaded into
> Maxima. At present the table is built by a Perl script.
> A question for consideration: building the look-up table
> with a separate script means there is another step in build
> process. But then there is no time spent building the table
> at run-time. That seems like an acceptable trade-off to me.
> Yes/no ?
> 

I don't see any problems with this.  Almost all help
systems have some kind of prebuilt indices.


-- 
      Vadim V. Zhytnikov

       <vvzhy at mail.ru>
      <vvzhy at netorn.ru>