I have some first-hand knowledge that could throw some light on this.
I was aware that Catalyst was being used within the iPlayer team, and
they were contributing many bug fixes back into Catalyst and DBIC. I was
not aware of this 'Perl on Rails' work. There again the BBC is big so it
is not surprising that one team does not always know what another team
is doing.
Catalyst is being used only for back end systems, not customer facing.
The BBC works almost exclusively on systems that publish HTML rather
than providing dynamic content.
J. Shirley wrote:
On Dec 1, 2007 11:29 AM, Jonathan Rockway <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
<snip>
I think that the 5.6 limitation was the main reason for not using
Catalyst.
The insanity of a few of their points lead me to believe their Perl on
Rails platform is probably not ever going to go anywhere:
- "we needed to expose any SQL queries we would make so they could be
vetted by DBA's for optimization"
- "On the live environment we were told at the time we had Perl 5.6,
and a few BBC approved perl modules."
I certainly believe that _some_ SQL queries should be optimized, or
rather the database optimized for those queries, making every query
that way is just madness. It certainly isn't extensible.
To support the BBC on this, the SQL needs to be made visible for
database optimisation, but mainly for the purpose of showing what
indexes should be put on the tables. Since it is possible to cause the
generated SQL to be output then this should be sufficient and is no
reason to avoid an ORM.
The best way to tie developers hands and waste money is by giving them
an immutable whitelist of modules. I wonder how many "utility"
methods are in this framework that should be broken out into separate
modules -- and in reality, there are probably already 3 modules
already doing their utility method. People really need to get over
NIH-Syndrome.
It is not a NIH Syndrome, the BBC live environment is supported and
maintained by Siemens. You would not believe the time and effort
involved in getting new modules approved and installed on the live
system (I am talking months if not years). Is it surprising the
developers lose patience and design their own ORM when they don't have
any other option? The DBAs and support teams are also frustrated at this
since it makes support so much more difficult having all these different
ways of doing the same thing.
Regards
Ian Docherty
_______________________________________________
List: [email protected]
Listinfo: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/[EMAIL PROTECTED]/
Dev site: http://dev.catalyst.perl.org/
