Want your opinion about a programming project

Thomas Baruchel thomas.baruchel at laposte.net
Wed Sep 17 14:45:10 CEST 2003


Hi,

as you know, Sloane's database can be downloaded as a gzip archive
and studied in a local way on your own computer. I think to a
tool that would be made in order to help at it by allowing highly
configurable way of searching the database. Please, read what follows,
and then tell me your opinion and suggestions.

Technical point of view:
========================
Program would certainly be written in C+ncurses+guile
Ncurses would provide a handy interface with menus, colors,
keyboard shortcuts, etc.
Guile would be the extension language in order to allow the
user to write his own definitions (like the ones superseeker
uses, but with as many extensions as wanted; for instance, I
often thought that it could be nice to look for a sequence l0
in the database such that each consecutive term of l0 is a divisor
of the respective term in l* (or is a multiple)).
Some options: uncompress the database at startup, uncompress it
and copy it in a RAM disk in order to increase speed and spare
the hard disk.
When looking at the result, shortcuts to make sendmail ask the
real database for informations about the sequence that was found.
Special interface (based on a FIFO file or something else) that
could allow your favorite mathematical software to send a list of
numbers to the program without complicated mouse copy/paste.
History mechanisms, diary allowing personal comments, etc.

Mathematical point of view:
===========================
The base of all would be guile. The program would read at
startup the configuration file of the user containing
definitions of his favorite search-extensions. It could be
the same than what superseeker does, but much more also.
An example is given above, but also convenient way to do
things like: look for l0 such that each term of l* is in
l0 in the same order, but allow "noise" between matching terms
of l0:
  lookup 1,2,3,*,17,19,*,241,*,17
  lookup 1,(3|6),7,*,11,(17|19)
  etc.
I know that you can do much by using grep, agrep, etc.
But these tools can't do everything (though I don't
exclude using myself regexp at a low level for some kinds
of job).
  
Purpose is not only to do the same job than superseeker in
a local way (which is already something), but also study
the database itself, by trying to find links between sequences.
For this purpose, guile would give to the user not only
a way of defining "transformations" before searching but
also the ability to write complex scripts.

I have not that much time, but if you think it is a good idea,
I would certainly take some time to do it. Please let me
kown if something has already be done in this way (maybe perl
or python scripts ?).

Cordially,

-- 
« nous devons agir comme si la chose qui peut-être ne sera pas devait
être » (Kant, Métaphysique des moeurs, doctrine du droit, II conclusion)

  Thomas Baruchel <baruchel at laposte.net>





More information about the SeqFan mailing list