[Biojava-dev] gaps and holes

[Matthew Pocock]

On Tuesday 01 November 2005 12:42, Kalle Näslund wrote:
> Well, atleast it sounds complex and neat =) Jokes aside, i can
> understand the need for
> two types of gap symbols, so i agree that we should keep them.

Hehe. It's a pita that it's needed, but I think it /realy/ is.

> I will try to read your mail, and the javadocs a few more times, until i
> think i understand
> how the symbol magic works, and i will then try to figure out where in
> the serialization
> stuff goes b0rk.

OK. I think the main magic would be to serialize things out as - or ~, and if 
you have a leading/trailing ~ then it may be reasonable to use {} rather than 
[] where as internal - is almost certainly meant to be []. Unfortunately, I 
think we may have lossy serialization here :( Would we need to pass some sort 
of 'how far through' hint to serialization? that feels sucky.

> So far it seems that if i take a small DNA SymbolList, 
> write it to disk
> and then read it. It no longer has the same symbols in it. So im fairly
> sure this realy is
> a serialization issue.

Yup. It would be key to find out exactly which gaps where written out and 
which where loaded in.

>
> /kalle

Matthew