
######################################################
# Less Urgent
######################################################

* simple user manual --> Always updating
   
    - asciidoc


######################################################
# To do
######################################################

* Make it faster
   The current implementation performs OK for chloroplasts genomes but
   it would be too slow for nuclear chromosomes.  In example, much of
   the geometry could be stored into OpenGL display lists instead of
   recomputed each time that we scroll.


* Suggest the hit quality
   there should not be too much arbitrary decision when telling if a
   marker is good or bad


* Support for per-probe comment


######################################################
# Not so urgent
######################################################

* Eject Needle and Blast?
   needle could be replaced by a standard dynamic programming algo
   (see Sylvie Hamel's thesis for an example (1st chapter))
   it could also use Sylvie Hamel's bit vector algo
   it could also use a hacked agrep that is not line oriented

* Support for per-user and per-machine config files

* integrate ICAass clustering

* Support for changing the BLAST parameters
   - word length 
   - masking
   - bl2seq vs blastall
   - ... 


######################################################
# Maybe one day
######################################################


* think big!
   split the align in chunks and compute the geometry of the curent
   block plus the geometry of its neighbors


* think about multiple chromosomes
   a tab for each chromosome? 
   compute all the 2 by 2 alignments ?
   compute the concatenated alignment ?
   ?????

* emphasis on portability
   does it work on Mac ?
   is it hard to make it work ?
   same for Windows


* autodetect beakpoints

* autodetect probes

