[ic] UTF-8 Advantages/Disadvantages
peter at pajamian.dhs.org
Sun May 30 01:50:05 UTC 2010
On 30/05/10 08:36, Chris Keane wrote:
> Heh, I understand the issue at a technical level I'm sure you can
> appreciate that from a release management perspective it's puzzling that
> an "upgrade" would slow down established code by a factor of 10.
My understanding of teh situation is that it is largely due to perl and
Encode issues. You may want to try installing perl 5.10.1 or even
5.12.1 and see if that helps (compile your own).
> Some of the more critical items in our system, such as the race gridding
> system (loops extensively) we've completely moved down into a usertag
> (which speeds the process from ~60 seconds per race on 5.7.2, ~8 seconds
> per race on 5.4, to ~2 seconds per race)
Is that down to ~2 seconds per race even on 5.7.2 with UTF8 turned on?
> However, I'm sure I don't have to mention how eye-rolling it is to have
> usertags that have HTML embedded in the code or loaded from our content
> management system rather than being able to use ITL.
> Makes me wonder at what point we should jettison Interchange altogether
> because all our important code is written straight in perl! I suspect we
> have one of the more customized and extensive systems written on
> Interchange in existence and "rewriting all our loops" will be
> problematic at best! We have our DB-backed CMS, CRM, kiosks, check-in
> desks, race control, results management, e-commerce, reporting and
> accounting systems, all ajax-ified. 7-8 years of development which would
> be painful to revisit completely, especially with no assurance that the
> "problem" can even be solved without rewriting in straight perl. I also
> don't want us to be stuck in 5.4 land, or non-UTF-8 forever since we're
> starting to sign non-english speaking clients.
> As a long time proponent of Interchange I think the IC dev group does a
> stellar job and I do understand the importance of UTF-8. I feel bad
> offering criticism when in this case I haven't gone so far as digging
> through the code myself to understand the challenges of the UTF-8
> integration. I suspect that more and more IC users will bump their heads
> into this one (look at Bill @ Bottlenose, for example), so if this was
> my project I would have this upward-revision performance disaster fairly
> high at the top of my critical problems list... but that's just me.
> Regardless, I love you guys and I offer this entire comment in a spirit
> of concerned interest.
Thank you, it is appreciated.
> If any of the core group need access to one of our test environments to
> mess around with let me know.
I wish I had the time, but right now paid work takes the priority.
Maybe someone else will take you up on the offer.
More information about the interchange-users