earle@MAHENDO.JPL.NASA.GOV (Greg Earle) (09/17/87)
I've just skimmed the `releasenotes' paper with X V11 R1. After noting all the bugs with the clients (`xxx dumps core' `xxx doesn't work on a color frame buffer' `xxx freaks out under wm' `xxx does this & that'); the fact that there is a `new' resource manager that will be the standard, but nothing uses it yet and uses the `old' one instead; and finally, noting the real long time allowed for beta-test Release 4 (all of a week and a half?), I'm wondering if someone didn't have a gun to Athena's (collective) head to get Release 1 out to the waiting public on Sept. 15th Or Else, no matter what the state of it was. Can any of the particulars comment? By the time I waded through the BUGS sections for all those clients, I began to wonder `Why release it if there are still this many outstanding (& recognized, that's really the key - you can't do much about the bugs you haven't found yet) problems'. I'm sure those in the know will have a better idea than I, but I just couldn't help being curious about that. Since this *is* the Official Godlike X Release, Standard Of The Known Universe, I would have thought that beta test would have lasted until even the niggly-ass leaks were spot-welded shut ... (then again, perhaps one should be taking the attitude that the Protocol and sample server are Truly Godlike, and anything else you get is gravy? - note that this is *not* a flame, I'm just wondering) Wait for Release 2, anyone? -- Greg Earle earle@mahendo.JPL.NASA.GOV Sun Consultant earle%mahendo@jpl-elroy.ARPA [aka:] Rockwell International earle%mahendo@elroy.JPL.NASA.GOV Seal Beach, CA ...!cit-vax!elroy!jplgodo!mahendo!earle
swick@ATHENA.MIT.EDU (Ralph R. Swick) (09/18/87)
Had your phones been as busy as ours the past few months, you might feel differently about our schedule. The figurative "gun" in this case was the outside world; we are being swamped with requests to "send everything you can". We would have liked very much to be able to take another month or two to clean things up, however in real time those 2 months would have required 6 while we responded to more requests for information/pre-releases. Somewhere in the editing process I had a paragraph about not expecting XV11R1 to be as shaken-out as XV10R4. The version 11 protocol is a significant change/improvement over the version 10 protocol. Re-shaping applications to take advantage of the new protocol is not an over-night process. We fully expect that it will take another release or two before all the new kinks are worked out. While we are not enthusiastic about some of the compromises we had to make (i.e. sending multiple versions of several libraries), we are confident that this release does meet the "useability" objective. We are also confident that the interfaces at the server and Xlib level, which are the critical ones for application portability, are robust and that it will be several years (cross your fingers) before another cataclysm is required. Were we a commercial organization instead of an academic one, we would not have slipped the release either; we simply wouldn't have told you about all the bugs. :-) As to waiting until Release 2 - it's your choice, but from the mail we've been getting, you'll be pretty lonely. -Ralph
RWS@ZERMATT.LCS.MIT.EDU (Robert Scheifler) (09/19/87)
If you look at the list of contributors, you will discover that V11R1 was a rather major public undertaking. Now, reflect on that, and realize that not all companies out there were contributing to the public effort, and that even those companies that were certainly didn't expend all of their energies in the public arena. Then think about a snowball rolling down a mountainside, and little MIT standing half way down thinking about whether to try and stop the snowball. It just wasn't sensible. There are enough companies out there, sinking enough time into it, fixing bugs and creating applications independent of MIT, and in some cases writing their entire systems from scratch. At some point we had to say "OK, you can go to product now". That time had come. The protocol and the major interfaces are in good shape. Sure, there are lots of rough edges on the code, and a fair number of stupid bugs. The stupic bugs will get fixed quickly, we'll let you know about them ASAP in most cases, and the rough edges will smooth in time. The next release, in December or January, will be a lot more polished. However, we were getting >1000 mail message a week, we had someone working full time just on trying to empty out someone else's mailbox, we were getting dozens of phone calls a day, people pleading with us for "bits, any bits", pleading for beta test access two days before the release, etc. Better to just get it out there. Better to have people complain that mumble dumps core on a color Sun, rather than have them continue to complain that X10 doesn't do splines on a Sun. This release is not Godlike; the protocol and library interfaces are hopefully like that, but the Sample Server should be viewed as just a sample, not as a Reference Implementation.