This is an automated notification sent by LCG Savannah.
It relates to:
task #7148, project CDS Invenio
==============================================================================
OVERVIEW of task #7148:
==============================================================================
URL:
<http://savannah.cern.ch/task/?7148>
Summary: make webcoll phase 2 fully optional
Project: CDS Invenio
Submitted by: simko
Submitted on: 2008-06-17 17:10
Should Start On: 2008-06-17 00:00
Should be Finished on: 2008-06-17 00:00
Category: WebSearch
Priority: 5 - Normal
Status: None
Privacy: Public
Percent Complete: 0%
Assigned to: None
Open/Closed: Open
Discussion Lock: Any
Effort: 0.00
_______________________________________________________
Currently webcoll its in phase 2 generates the search interface web cache for
all collections and all languages. This may be slow because it needs to
generate the list of latest additions in every language. Moreover, when the
web cache is not generated, the search interface displays only an error which
is not very nice even though it can occur in practice rarely, only when one
defines a new collection and webcoll has not been run yet.
What would be nice is to make the search interface to use the webcoll web
cache fully optionally, like RSS or WebJournal use their caches. If the
cached files are there, use them; otherwise generate them on the fly.
In this way the webcoll daemon could do only the part 1 of its job (=creation
of the record universe for every collection) which is necessary for the
virtual collections. This would permit us to run it much more frequently.
The price to pay for this change is that the web cache would not be
precreated for all languages so when a user of a lightly used language hits
the web site she might be the unlucky one that has to wait for the generation
of the 'latest additions' box which may take a couple of seconds for big
collections such as preprints with its 400k of records. But this seems
acceptable as it should not occur often.
In summary, a lazy generation of the collection web cache is required, while
the webcoll daemon would be pre-generating the reclist universe only.
_______________________________________________________
Carbon-Copy List:
CC Address | Comment
------------------------------------+-----------------------------
1576 | -SUB-
==============================================================================
This item URL is:
<http://savannah.cern.ch/task/?7148>
_______________________________________________
Message sent via/by LCG Savannah
http://savannah.cern.ch/
