On Mon, 2003-04-28 at 05:03, Björn Stenberg wrote: > Matthew Palmer wrote: > > I don't think your proposals will really fix that, since in my experience > > that new version of A probably requires all sorts of new crap from B > > anyway... > > Does it, really? Or does it simply have binary dependencies to an unstable > version of B, imposed by B? If one version of A and B has been accepted in > testing, chances are pretty good that the next version of A will compile and > work fine against that old version of B. Most new versions are fixes, either > of bugs or features. Changes that break source level compatibility are rather > rare.
Well, I guess you haven't been burned yet, then. Anybody who has worked for any significant amount of time in quality control knows that there is no substitute for testing in the *exact* environment that you plan to release in. Period. Let me give you a sort-of off-topic example. I'm building a self-contained system for demo purposes, built on top of Debian and a bunch of other crap. The company purchased a Toshiba laptop (not really what I wanted) because it was cheap and sold and serviced locally, which has a GeForce4Go video system. I installed (among other things) the nvidia drivers and openoffice (they want to show PowerPoint slides at the same time and don't want to carry two laptops); I also tested the whole system and verified that it worked correctly with a projector attached, but I didn't have the same model of projector attached. When the whole pile of gear was taken over to the customer's site for a demo, they had a different model of projector there, but that shouldn't be a problem, right? Wrong. TrueType fonts don't display as long as that projector is connected. I have no idea why, I just know that they don't display as long as that projector is connected. And now you want to tell me that I shouldn't be concerned about some "minor changes" in a library, and besides, it should still be API-compatible and therefore it will work? Sorry, but I just don't believe that. There is no such thing as "innocent until proven guilty" wrt system changes. *All* changes are suspect until they have been *proven* to work, and that includes all of your little "but I didn't change anything important" library updates. The point is that we're trying for more than just "source level compatibility". "It compiled" is *not* the gold standard in quality. -- Stephen Ryan Debian Linux 3.0 Technology Coordinator Center for Educational Outcomes at Dartmouth College
