[EP-tech] Revision cleaning system

[Paolo Tealdi]

Dear all,

trying to use the RemoveRevisions plugin (this one : 
http://files.eprints.org/615/ ) i noticed that something was not going 
as expected.

After a lot of debugging i saw that in my 3.2.4 eprint repository 
purging an eprint didn't clean the revision  records in the file table : 
for each eprint now i have toomany records in file table (many pointing 
to old files and some to the correct new).

This effect is very big in my repository  because it is readonly : all 
records are batch imported, and  before importing every eprint is 
deleted (if present)  and freshly imported as new. This action erases 
the xml revision files but doesn't delete the appropriated record in the 
file table.

I noticed it because i'm testing to migrate to 3.3.10 where i changed 
the importing batch, using the --update function of import file.

To resolve the proliferation of such records in file table i modified 
the RemoveRevisions plugin to check and possibly remove all those 
records in file table that are inconsistent  with  xml revision files .

After having copied the attach in 
~/archives/"repository"/cfg/plugins/EPrints/Plugin/Export directory if 
you execute

~/bin/export "repository" --verbose archive CheckRevisions

you can do a  check only. If you use the --arg write=1 you delete 
inconsistent record in file table, without deleting files in the 
revision directory.

Any comment is very welcomed.


Best regards,
Paolo Tealdi

-- 
Ing. Paolo Tealdi         Area IT - Politecnico Torino
Telefono/Phone : +39-011-0906714 , FAX : +39-011-0906799
Indirizzo/Address : C.so Duca degli Abruzzi,  24 - 10129 Torino - ITALY
Skype : tealdi.paolo
Please consider your environmental responsibility before printing this e-mail

-------------- next part --------------
A non-text attachment was scrubbed...
Name: CheckRevisions.pm
Type: application/x-perl
Size: 2428 bytes
Desc: not available
Url : http://mailman.ecs.soton.ac.uk/pipermail/eprints-tech/attachments/20120927/c968bb92/attachment.bin