Testing is testing means go to "most" software company and ask to see their test procedures and then compare those to what you are likely to find in many open source projects ;-). You hit it on the head with the notion that companies slack off on testing. Which incidentally was the entire point of being sick of the Fedora debate. Good points. Wade -----Original Message----- From: redhat-list-admin@xxxxxxxxxx [mailto:redhat-list-admin@xxxxxxxxxx] On Behalf Of Dave Ihnat Sent: Tuesday, November 04, 2003 8:38 AM To: redhat-list@xxxxxxxxxx Subject: Re: Anyone planning to use Fedora in production? On Tue, Nov 04, 2003 at 07:55:52AM -0500, Wade Chandler wrote: > Geesh. Who in here has worked at a software company before? Testing > is testing. Uh...I've set up software company testing before. "Testing is testing" is only for amateur efforts. REAL testing involves regression tests to guarantee nothing that did work is broken and branch testing to assure that nothing you wanted to include was omitted, *AND* that nothing you DIDN'T intend to include was added (that *is* important--anything added permutes existing code). This usually means automated testing with regression suites and reporting/replay functions. The test LAN should be a dedicated, isolated environment. There should be multiple systems covering the architectures you're going to claim you support. Now, granted that all this changes to a degree for an operating system. Moreso for Open Source projects incorporating multiple complex packages. But the fact is that for a vendor to claim a product is tested, some degree of rigor is necessary. I don't know if RedHat ever subjected its releases to any such degree of testing, and I don't know if Fedora will undergo any formal testing by RedHat personnel. But I do know that the informal, catch-as-catch-can testing carried out after release--even by "many eyes"-- will almost certainly miss bywater errors. (Why did I post this? Partly because so many people short-shrift formalizing testing efforts, making it hellishly difficult to get equipment and software approvals to set up real testing labs. And I've gotten sick and tired of the slipshod, buggy, "release mega new features by THIS date, cut the testing if necessary" attitude of many software vendors (Over the last 20 years or so--it seems, FTM, ever since a certain Redmond company came on the scene, but maybe that's pre-caffeine choler...) I guess I want people to know there IS a better way, and maybe argue for it when it's appropriate. -- Dave Ihnat ignatz@xxxxxxxxxx -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
