"Tim Churches" <tchur@BIGPOND.COM> wrote:

>Seriously, the fact that SI has taken so long to deliver a production >release of SAS Version 8 despite apparently having plenty of resources >(in terms of revenue and employees) at its disposal does suggest a few >possibilities: a) there is something seriously wrong with the way in >which SAS Institute is organised which leads to low productivity >(unlikely) ; b) far too many of those 5500 employes are engaged in >marketing instead of software engineering (a distinct possibility);

I don't think so. I haven't seen any indications that the technical side of SAS Institute is underfunded for any product that the market cares about (I make that exception because OpenVMS support is not what I think it should be - lots of important things aren't in v7/v8 - but I realize that the market, as a whole, has abandoned OpenVMS).

Two things to keep in mind:

- Even if SAS Institute wanted to throw another 10,000 programmers at the code, it wouldn't do any immediate good. The system is just too big for new people to be able to make worthwhile immediate contributions which fit in well with everything else.

- The marketing people get more publicity than their numbers might justify because, well, because marketing people are paid to get publicity. I don't know how big the marketing and publicity departments are at SAS Institute, but I'd be surprised if they take up a larger percentage of the budget than the equivalent departments at other companies of similar size.

>c) SI is so obsesssed with quality assurance that it is unable to bring >itself to ship a new product (unlikely because although previous SAS >releases have been relatively bug free, they have not been faultless) or

heh. They do seem to be more concerned than they have been in some previous releases. I remember an MVS installation tape that had a JCL error in the very first job, so not even the installation program could be installed - obviously no testing had been done. I haven't heard of anything that bad coming from SAS recently.

>d) the underlying source code of the SAS system is such a mess that it >really has taken lots of programmmers over three years to incorporate >the relatively modest extensions introduced in Version 8.

Given how much was left out of v7, I'm not sure that the extensions are "relatively modest". My guess is that at least a few medium to large size chunks were reworked between v7 and v8.

>Because SAS is >closed-source software, it is impossible to comment on this possibility.

Not impossible. We can get an idea by looking at the internals which are documented - SAS/Toolkit, for example, when it comes out. We can see whether the language and procedure extensions are complete and logical. We can see whether the errors in the software are of a type which indicates sloppy coding. I think it's too soon to tell.

-- JackHamilton@FirstHealth.com Development Manager, Technical Group METRICS Department, First Health West Sacramento, California USA