Re: Informix 7.3 on Linux -25580 Error

"Obnoxio The Clown" <obnoxio@hotmail.com> writes: > From: Thomas Parsli <thomas.parsli@startsiden.no> > > > >dwight@rudolph.com (Dwight Rudolph Jr) writes: > > > > > In article <m3em5ocrqi.fsf@yakisoba.forte-intl.com>, > >ronald@forte-intl.com > > > says... > > > > > > > > > > I'm just a bit reluctant to throw more money at this problem, > >considering that > > > 1) it's MY money, and > > > >It's bloody inexpensive, you know that? > > I'll take your word for it. :0) Take Intrawares word for it: "Promotionally priced at $199 per user and $2000 per CPU!" > >Sorting is the only ting I _really_ hate on Linux, probably because > >of cooked-files, dbspaces for sorting took about three times as long as > >using filesystem (haven't played with raw-patches or using devices as > >files). > > Tsk. Tsk. Have you tried PSORT_NPROCS and (more importantly) PSORT_DBTEMP, > then? Yes, but probably not enough... I'm (still) waiting for 2.4 to emerge as the holy penguin -with raw disk, and a better scheduler;) > >I hasn't crashed (yet), receives ~220.000 updates each day (small records > >in that table) and generally seems just fine. > > Must have slipped past QA, then. Whaddayamean? QA? -Didn't know they did that anymore... > Personally, I think it's important for people to get over the fear of 9.x > hurdle. There's a whole new world on the other side, provided you ignore the > DataBlade BS. I mean, DataBlades may help a lot of people, but not the > majority (IMHO). I think datablades are great, the lack of "good ones" are not that great -but time will probably help. > But the ability to extend the database and embed real > business-valuable login into the server is a cracker. The new datatypes and the possibility to extent those are great, but we still can't create indexes on those -and forget about foreign keys... > I'm working on a data warehouse with IDS.2000 and there are so many > constraints I've been freed from because I don't have to limit myself to > "conventional" database technology. My C is taking a beating, because it's > so rusty, but as I move along I can definitely see that Informix has once > again been beating entirely the wrong drum and mismarketing IDS.2000 (and > IUS 9.x, by implication) The have actually done marketing? Informix is still a database little known, with great features even lesser known... > -- forget DataBlades, think about the perks of > being able to write business logic or components directly into the server in > a real language, not that abomination called SPL. Don't have quite the right > datatype available? Write your own. Etc. If they actually _had_ a marketing dept.... Thomas

Thomas Parsli wrote: > The new datatypes and the possibility to extent those are great, but > we still can't create indexes on those -and forget about foreign keys... Woah nellie! You can't index ROW TYPES (yet). But you sure as shoot can create indices on OPAQUE and DISTINCT types (that's how the R-Tree stuff for spatial works), defined primary keys using them, and also referential integrity constraints. And if you *really* want to see the power of this stuff, have a quick look at the "Managing Hierarchical Data" example on http://examples.informix.com/. If you're managing a parts explosion problem that's really big, nothing can beat this. (http://examples.informix.com, then "Using DataBlade Modules", then "Manage Hierarchical Data" at the bottom of the page).