info-vax@ucbvax.ARPA (03/11/85)
From: sasaki@harvard.ARPA This is a mild flame. If you agree with me, let your DEC salesman know that there is a need, and that you are willing to pay a rea- sonable sum to fill this need. Mail to non-VMS machines is very important to many sites (ours included). The many work-arounds in existence point out this need, yet DEC provides *NO* support for this. If you don't use DECNET, then you lose. There is no externally available documentation for the Mail-11 protocol. There is no documentation for the various hooks in the mail system. The only source for these things is the micro-fiche, and the documentation provided there is also lacking. The only way that I can send mail to a TCP/IP site (using DEC "supported" software) is to route the mail through an ULTRIX site running the officially un-announced DECNET-on-ULTRIX. Inquiries about future plans for TCP/IP support on VMS (even with non-disclosure) have yielded only blank stares. DEC is developing (currently in field test, I think) a "smart" ethernet controller. The only protocol that is being supported is DECNET. This total lack of support has lead to the various work-arounds. A few sites have completely abandoned using DEC supplied mail and are using various home-grown systems, most based on SMTP. It is time for DEC to wake up and provide some kind of support for network operations via non-DECNET protocols. The company line about official non-support, and "your code will break in a future release" really isn't acceptable. Marty Sasaki Havard University Science Center sasaki@harvard.{arpa,uucp} 617-495-1270
info-vax@ucbvax.ARPA (03/13/85)
From: medin@ucbarpa (Milo Medin) Did you really expect anything different? The bigger they are the less responsive they are. I personally wwouldn't like to see vaxmail kludged to run SMTP. Now, TWG makes a mailer patch for their TCP/IP stuff, but its ugly, and reply doesnt work right. It cant, not without messing with vaxmail itself. And why would you want to use vaxmail? No store and forward, no aliases, no cc, etc... Its just like that Chevron commercial, are you tired of being said no too? Now, I understand Dave Kashtan is working on porting a version of MM to VMS. Tie that in with SMTP and you have a real mail system, not a toy like vaxmail. I agree that people don't want to learn more than one mailer, but I don't agree that mailer should be vaxmail. Anyone who has ever used Unix mail (even v7) or TOPS-20 MM gets frustrated with vaxmail. And the Rand MH package makes Unix mail even better. But then there is the issue of software support, and that has been raised before. I'll take a widely distributed piece of useful code (with source) over a supported piece of non-useful code any day. Noone supports 4.2 per se, but its alive and well today,and its because its fixable. If I find a bug (esp. a security bug), I want it fixed NOW, even if I have to do it, or get someone else to do it, rather than fill out an SPR and wait 6 mons. for a new release of the software. Even then, DEC may not have fixed it the right way. I deal with both 4.2 and VMS machines all the time, and I generally deal with mail on my 4.2 machines. The reason I don't do it all the time, is that there is no aliasing or automatic mail forwarding in Vaxmail. The TWG stuff does implement aliases though, but only for messages coming in with SMTP. If DEC wants to do the right thing (listening DEC?), port the Rand Mailer over.... And don't rewrite it in bliss.... Milo
info-vax@ucbvax.ARPA (03/14/85)
From: amdcad!phil@BERKELEY (Phil Ngai) I would like to address this "bigger they are the less responsive they are" propaganda. The bigger they are, the more problems they have to solve. They have an incentive to come up with one all encompassing solution rather than little ad-hoc solutions here and there. If they didn't, then they would be flamed about how all their stuff doesn't play together. A company that makes N different systems doesn't have N times as much support, they have N^2 times. Sure a motorcycle can beat a train for delivering one package. But try sending 500 tons of coal by motorcycle. It is hard for me to believe I am defending DEC when our VAX sat, non functional, on our computer room floor for two months because they couldn't make it work. But here it is.
info-vax@ucbvax.ARPA (03/14/85)
From: (Stephen Tihor) <TIHOR@NYU-CMCL1.ARPA> Sadly though there are not too many places you can buy well supported 4.2 SYSTEMS from yet. We have several here at NYU and other than Mt. Xinu and the hardware vendors no one was able to suggest a support alternative better than "Well go and hire a Unix Wizard". Now the hardware vendors are companies small and large but as long as you code is portable it can be moved. Sadly, for our scientific users, few of the large numerical codes and libraries are written in C. Few vendors are really good fortran compilers under 4.2. The few than do usually don't have IP/TCP working right. The last couple of systems are almost livable they'll get our business based on price/perfomance and maintanence considerations. Unlike some sites we don;t have that many problems with DEC FS. Sometimes you have to train a new CE on how to install a new kind of system or what to check. But whenever we get a real bozo we just recycle him and DEC send us someone else. So we have three times as many cycles of VAXen...and five times as many usable cycles. \\ Stephen Tihor / CIMS / NYU / 251 Mercer Street / New York, NY 10012 // (( DEC Enet: RHEA::DECWRL::"""TIHOR@NYU-CMCL1.ARPA""" NYUnet: TIHOR.CMCL1 )) // ARPAnet: Tihor@NYU-CMCL1 UUCPnet address: ...!ihnp4!cmcl2!cmcl1!tihor \\ -------
info-vax@ucbvax.ARPA (03/15/85)
From: medin@ucbarpa (Milo Medin) I agree with you about the lousy f77 compiler, its fairly pitiful. Someone wanting to do good for 4.2 should write a good f77 compiler for it. But if you need fortran, VMS is excellent for that. I am biased but not irratiional; I recommended to a Math Dept. a while back that they convert one or both of their 750's to VMS to run FORTRAN. But they didn't want to, because the user interface is so poor, anyways, thats what they said. I guess I have been too hard on DEC in this list, but I'd like to say that I'll take DEC over IBM or CDC or DG anyday. I've talked to many folks who work in DEC research, and they have some truly incredible stuff. And they know what they are doing too. But the marketing types keep getting in the way, and comprimising performance for various reasons. The same is true for IBM, I understand they had a relatiional database for years before they released it, because their old database stuff was selling well. And DEC is probably slow in coming out with the Microvax II because it would destroy the 780-750-730-725 market, and take a big chunk out of the 785 market as well. And no cluster controller for it yet... At least, thats true if the Microvax II is as good as the rumours indicate. Milo
info-vax@ucbvax.ARPA (03/16/85)
From: medin@ucbarpa (Milo Medin) All you say is true, but to the people they sell to, that is quite irrelevent. If I am getting lousy service (and I do) then I really want a solution, not an excuse. If any BIG company cant fill my needs, then I won't go to a BIG company for my business. I have had Vaxen wait for months to be installed, software not shipped in the allowable time for that contract, and once it was shipped, it wasn't the right software. Eventually, I was told to copy it from someone else because we needed it quickly. This didn't happen at UCB by the way, but another place. In any case, there are many places to buy 4.2 machines from, of course if you run VMS, well.... Milo