IUBio

[Re: indexing subdirectiories]

José R. Valverde txomsy at cnb.uam.es
Thu Jul 3 04:33:20 EST 1997


In article <33B93604.B28DA1C at ebi.ac.uk>,
	etzold <etzold at ebi.ac.uk> writes:
> Mike Tennant wrote:
>> 
>> Hi,
>>         I'm trying to index our pdb hierarchy, which has the format
>> /pdb/dist/*/*.ent.gz, but can't get srs5.05 to get past the first wild
>> card (which is a set of directories). Does anybody know whether this can
>> be done?
> 
> funny! i am working on that right now...which means that it cannot 
> be done at the moment. This will be possible with the next release
> 5.0.6. 
> 

    Well, meanwhile, maybe it would work to make links to those files in
another directory.

    Mike doesn't say, but I guess the problem he refers to is that he's
got the entries sorted by starting letters in corresponding characters.
Now, if SRS wants them in a single directory, a paliative solution
could be to link the files in a single directory. Something like

	cd /.../pdb/dist
	mkdir all_entries
	cd all_entries
	ln -s ../*/* .

    You may even remove the "s" and make those hard-links, wherever the
files would be exactly the same as the original ones, only they would not
take any more space than the directory entries take, which should be a
lot less than the actual files themselves.

    Of course, you may have some trouble if you update PDB nightly, since
you'd already have thousands of old links, and running that nightly would
waste a lot of cycles and create many errors. Nothing easier to fix: use
'find' testing for '-ctime -nodays' where 'nodays' is the period with which
you update your links.

    I know, it's a kludge, but works and saves a lot of space and trouble.
After all, that's one of the beauties of Unix, isn't it?

				jr

-- 
Jose R. Valverde
EMBnet/CNB




More information about the Bio-srs mailing list

Send comments to us at biosci-help [At] net.bio.net