Re: Informix 4.0 (was Re: Ancient History)

Path: emory!swrinde!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!rutgers!pyrnj!pyramid!infmx!aland From: aland@informix.com (Colonel Panic) Newsgroups: comp.databases.informix Message-ID: <1991Sep28.050032.326@informix.com> Date: 28 Sep 91 05:00:32 GMT References: <8269@ns-mx.uiowa.edu> <1991Sep25.043919.25626@informix.com> <8302@ns-mx.uiowa.edu> Sender: news@informix.com (Usenet News) Organization: International Brotherhood of Geeks, Local 619 In article <8302@ns-mx.uiowa.edu> lbrintle@umaxc.weeg.uiowa.edu (Lee Brintle) writes: >In article <1991Sep25.043919.25626@informix.com> aland@informix.com (Colonel >Panic) writes: >>In article <8269@ns-mx.uiowa.edu> lbrintle@umaxc.weeg.uiowa.edu (Lee Brintle) >writes: >>>We've had a *lot* of problems with Informix (SE) 4.0; for example, one cannot >>>compile *any* runner under the RDS system for our platform. It appears that >>>they did not even try to run their product before releasing it... >> >>what platform is this? what problems do you get compiling new runners? > >SCO Xenix running on a Intel 386 box. Pretty generic configuration; >SCO/Xenix is a popular (I think largest) version of Unix and Intel/386 is not >exactly an unheard-of box. The compiler script is busted. As you mention later >on, some doof put the libraries in the wrong order. May be a different problem. I looked thru your cases, and the described symptoms from back then are different; yours wasn't indicated to be an undefined symbols problem. As far as I can tell, they story was that the test case was missing a "return 0" to indicate no values were returned (the example in chapter 1 omits the return statement, which can be misleading, but the examples in chapter 2 and in _4GL_By_Example_ have it correctly). Or, maybe this isn't the same case anyway... if not, can you recall the rough date? >Shipping a shell script that crashes every time it's run? I'm sure that says >wonders for whatever testing suite you use. Again, assuming we're talking about the unresolveds problem... Most of the affected environments have global default CC argument files (e.g. /etc/default/cc), where if you stick an option in there, you get it for every compile. If the porting agent isn't careful to keep that at the system default, the test suites can all pass in spite of the omission. >Lee Brintle | ``And so, I leave you with this final word: >Leepfrog / Tanj | twang.'' -- Alan Denney aland@informix.com {pyramid|uunet}!infmx!aland "Feel free to hit 15 pound jackrabbits with your Saturn." - Charles Guest