Not really meant as overall arguments against browse, just pointing out
two current issues with Browse:
pagination counts incorrect: https://jira.duraspace.org/browse/DS-954
I'm not sure what is going on there, but a couple of things immediately
leap out:
1) The sort_value should be unique (for author, subject, etc. browses)
in the table - i.e. there should only be one display value in the table
(if two different "values" create the same normalization, then just one
should be used).
2) If DISTINCTs are being used in the [read] queries, then that is a
problem - the design was meant to eliminate that, as that is costly at
lookup time [for a large repository].
>From a user experience perspective, I think it wouldn't hurt if we
look at the browse interfaces & discovery interfaces, keep what works
and get rid of clutter/things that don't work.
From a systems perspective, I see no reason why both the browse indexes
and discovery indexes should be kept on the mid or longer term. But I'm
saying this without detailed knowledge about how they are different.
But it seems they're both consuming resources & basically offering the
same kind of services.
Clearly Discovery indexes can easily show a browse list of items (i.e. a
search for "everything"), and can sort (and filter(!)) it in various
ways.
What I'm not so clear about is the capability / usability of Discovery
to perform the function provided by the author, subject, etc. browse
lists - to show a list of all of the distinct terms for a particular
field, and from there browse the items.
In theory, these could/should(/do) operate as facets in the Discovery,
but there may be a usability issue when the number of possible values
for a given facet may be large (e.g. thousands of authors)?
From a resource perspective (and providing the implementation is correct
- i.e. avoiding DISTINCT lookups), the "legacy" browse is pretty light
from a resource perspective - it's just a small amount of additional
tables inside an existing RDBMS instance.
Regards,
G
------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Dspace-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dspace-devel
