daemon@decwrl.UUCP (02/06/84)
From: atfab::wyman From: ATFAB::WYMAN I think it would be a very unfortunate mistake if domains on the USENET were implemented along the lines of any of the "Area" specification methods suggested (such as Telco Area Code). To a great extent, one of the real benifits of the Domain concept is that it provides for the "logical" structuring of the network, as opposed to the purely physical structure which currently is forced by the requirement for explicit path name specification. The individual sender of a mail message should not be concerned with the network topology, nor should that user be required to analyze Telco tariffs to determine the most economical routing to a particular node... If someone is trying to get a message to me at DEC they will hopefully be able to rely on the domain structure to ensure that it gets here quickly and cheaply. bob wyman
mc68020@gilbbs.UUCP (Thomas J Keller) (09/07/86)
All this talk of domains, smart mailers, etc. is fascinating, but I
wonder if all you high-powered gurus have stopped to think about how many
of us out here do not have source licenses, and thus cannot blithely
modify our mailers and uucp systems? How many of you even give a damn?
(I should waste my time asking *THAT* question?!!)
That's all right. Obviously, only wealthy sites with source licenses
should be on the net anyway, right folks?
--
Disclaimer: Disclaimer? DISCLAIMER!? I don't need no stinking DISCLAIMER!!!
tom keller "She's alive, ALIVE!"
{ihnp4, dual}!ptsfa!gilbbs!mc68020
(* we may not be big, but we're small! *)ahby@meccts.UUCP (Shane P. McCarron) (09/08/86)
In article <895@gilbbs.UUCP> mc68020@gilbbs.UUCP (Thomas J Keller) writes: > > All this talk of domains, smart mailers, etc. is fascinating, but I >wonder if all you high-powered gurus have stopped to think about how many >of us out here do not have source licenses, and thus cannot blithely >modify our mailers and uucp systems? How many of you even give a damn? >(I should waste my time asking *THAT* question?!!) > > That's all right. Obviously, only wealthy sites with source licenses >should be on the net anyway, right folks? Don't be an idiot... Or at least, don't let your ignornace show quite so well and claim to be the injured party. You don't need a source license, or even sendmail to use domain style addressing. There was a source sent out just recently in mod.sources which was a replacement for rmail. The program (actually, set of programs) allows you to use full domain addressing, as well as smart rerouting of mail coming out of and through your site. Next time, think and look a little before you snipe at the source license sites who brought you the utilities you need to live in this community! -- Shane P. McCarron UUCP ihnp4!meccts!ahby MECC Technical Services ATT (612) 481-3589 "Sinners can repent, but stupid is forever."
guy@sun.UUCP (09/08/86)
> All this talk of domains, smart mailers, etc. is fascinating, but I > wonder if all you high-powered gurus have stopped to think about how many > of us out here do not have source licenses, and thus cannot blithely > modify our mailers and uucp systems? From my quick look at the information files that come with the "smail" mailer (available from the "mod.sources" archive), the "high-powered gurus" do seem to have thought about that. It comes with a "front-end" to "/bin/mail"; you move "/bin/mail" to "/bin/lmail", and the new "/bin/mail" will either runs "/bin/lmail" if you are asking it to read your mail, or "/bin/rmail" (a link to "smail") if you are asking it to send mail. It doesn't seem to require any changes to UUCP, and it doesn't require you to have "sendmail", although it will work with "sendmail". A little less flamage would seem to be in order. -- Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.com (or guy@sun.arpa)
stuart@BMS-AT.UUCP (Stuart D. Gathman) (09/10/86)
The names used on the network should be the most efficient for the
network. Making things easier for the user is the job of the software
on the local machine. Although I am not sure I understand domains,
I am very much in favor if they simplify routing. Aliases work fine
for frequent correspondence. The only thing I ask is that my mailer
be able to reply to a message using the information in the header.
Having said that: what do we need to stay compatible with the network?
We currently have SysV mail and uucp with the news software.
I mailed a request for 'pathalias' source, but it didn't make it
because a machine en route did not allow files over 50K. I have
sent another request, but nothing has come in yet. Evidently,
pathalias supports the old system, we now need 'smail'. Is there
a document which describes the new design and what software
modules are required? Given some specs, I could even write
some of the modules myself if they are not available.
Perhaps the things I ask are on the net somewhere. Please be
patient and just E-mail a response.
P.S. Our SysV mailer just copies the domains verbatim from
path lines for responses. Sometimes this works; sometimes I have to strip
the domain addressing for the mail to go through. Is this
because some machine en route is fuddy duddy?
--
Stuart D. Gathman <..!seismo!{vrdxhq|dgis}!BMS-AT!stuart>mc68020@gilbbs.UUCP (Thomas J Keller) (09/11/86)
In article <522@meccts.UUCP>, ahby@meccts.UUCP (Shane P. McCarron) writes: > Don't be an idiot... Or at least, don't let your ignornace show quite > so well and claim to be the injured party. You don't need a source > license, or even sendmail to use domain style addressing. There was > a source sent out just recently in mod.sources which was a replacement > for rmail. The program (actually, set of programs) allows you to use > full domain addressing, as well as smart rerouting of mail coming out of > and through your site. > > Next time, think and look a little before you snipe at the source > license sites who brought you the utilities you need to live in this OK. Here we go again. Listen, friends, I aprreciate the time and trouble that many of you go to to provide software and guidance (not to mention communications and maintenance) for the net. I recognize that it is virtually impossible to write software that will port directly to *EVERY* version of UNIX that is out there on the net. *HOWEVER*: the use of long identifiers (in both C code AND preprocessor code), and other relatively recent additions to C makes it EXTREMELY difficult for many of us to actually USE that software. Some of us can't *GET* other C compilers, for various reasons. Some of us are/were unfortunate enough to have purchased systems from vendors who don't give a flying f**k, and we can't afford to scrap them and buy new ones. The new mailers make heavy use of neat goodies like enumerated data types, which many of us pcc based users can't deal with. Which puts us right back where we were, doesn't it: if you're not wealthy enough to have the latest and greatest, screw you! (I might also add that while I can expect, and to some extent am willing to accept ad hominem attacks in places like net.politics or net.women, I do not consider them in the least acceptable or appropriate for this group. ) Furthermore, I have received several letters with much the same flavor as this. What is it with UNIX people, does the software interfere with your ability to treat someone who isn't as knowledgable with a little dignity and respect? To accept that not everybody knows everything there is to know about UNIX, and that this does *NOT* make them worthless or pitiful? It is this sort of "maintain the mystery" attitude among UNIX types which is primarily responsible for the difficulty UNIX is having breaking into the popular computing market. Keep it up, and you just might succeed in killing` UNIX off! -- Disclaimer: Disclaimer? DISCLAIMER!? I don't need no stinking DISCLAIMER!!! tom keller "She's alive, ALIVE!" {ihnp4, dual}!ptsfa!gilbbs!mc68020 (* we may not be big, but we're small! *)
david@ukma.uky.csnet (David Herron, NPR Lover) (09/11/86)
In article <895@gilbbs.UUCP> mc68020@gilbbs.UUCP (Thomas J Keller) writes: > All this talk of domains, smart mailers, etc. is fascinating, but I >wonder if all you high-powered gurus have stopped to think about how many >of us out here do not have source licenses, and thus cannot blithely >modify our mailers and uucp systems? How many of you even give a damn? >(I should waste my time asking *THAT* question?!!) Oh lord. Some of these programs can be tossed in the middle of existing binary systems. (I think I'm thinking of the "uumail" thing that was posted awhile back, or maybe I'm thinking of smail... no matter, the idea is still valid anyway). What you do is toss out that silly old /bin/rmail which came with your machine and insert one that recognizes domains. Then at your other end you have a user interface for sending mail. You can write your own if you like (or take "elm" or there was a copy of mailx posted some time ago or I think mh is probably easily enough available...). Or if you don't like to use your own user interface you can insert a "fake-uux" in /usr/bin which looks for rmail executions like: uux - remote-host!rmail domain!user (or some such...) and takes the path to remote-host, takes the first element of that and gives that to uux and gives the rest of the path as the argument to the remote rmail, i.e. uux - neighbor!rmail path-to-remote-host!domain!user In other words, this nice wonderful net CAN give you domain-ist-ing WITHOUT requireing source mods... even if it IS a silly hack. :-) -- David Herron, cbosgd!ukma!david, david@UKMA.BITNET, david@ms.uky.csnet (I'm also "postmaster" at all those addresses) (And "news" and "netnews" and "uucp" and ....)
guy@sun.uucp (Guy Harris) (09/12/86)
> *HOWEVER*: the use of long identifiers (in both C code AND preprocessor > code), and other relatively recent additions to C makes it EXTREMELY > difficult for many of us to actually USE that software. Does "smail" use long identifiers not unique in the first 7 characters? If not, no problem. If so, you *can* get the source to "smail", and fix it. Admittedly, this isn't optimal, but it doesn't leave you totally out in the cold. If you *do* have to fix it, and do so, why not send it back to the authors so that future versions will work on such machines? > The new mailers make heavy use of neat goodies like enumerated data > types, which many of us pcc based users can't deal with. PCC-based compilers have no problems with enumerated data types. They were added around 1979, and were present in the Ritchie C compiler distributed with V7. The document "A Tour Through the Portable C Compiler" indicates that the PCC distributed with V7 also had enumerated data types. Perhaps you have a *very* old compiler, in which case you have my sympathy, but it's not clear that writing "smail" to work with compilers *that* old doesn't go past the point of diminishing returns. I know of only one compiler that claimed to be V7-vintage that didn't support them, and it's long gone. > Which puts us right back where we were, doesn't it: if you're not > wealthy enough to have the latest and greatest, screw you! No, it doesn't. At least in the case of "enum"s, you hardly need the "latest and greatest". If it uses long identifiers not unique in the first 7 characters, I can understand why they decided to do this, but I think it was probably the wrong thing to do, since compilers that don't support them are still present on a lot of machines and are likely to remain on those machines (PDP-11, UNIX PC, etc.) for quite a while. > Furthermore, I have received several letters with much the same flavor as > this. What is it with UNIX people, does the software interfere with your > ability to treat someone who isn't as knowledgable with a little dignity > and respect? No, but uninformed flames *do* tend to interfere with that ability. If somebody who isn't knowledgable says "XXX is stupid" without knowing some of the reasons why XXX was done, and why *not* doing XXX might have been worse, patience tends to evaporate. > It is this sort of "maintain the mystery" attitude among UNIX types which > is primarily responsible for the difficulty UNIX is having breaking into the > popular computing market. Have you done research to support this claim? I suspect it has more to do with: 1) the high cost of UNIX systems relative to IBM PCs and the like 2) the relative lack of application software for UNIX 3) the fact that few, if any, UNIX machines are also IBM PC compatible machines As I said, I think a little less flaming is in order in this case. -- Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.com (or guy@sun.arpa)
mark@cbosgd.UUCP (Mark Horton) (09/14/86)
In article <901@gilbbs.UUCP> mc68020@gilbbs.UUCP (Thomas J Keller) writes: > *HOWEVER*: the use of long identifiers (in both C code AND preprocessor >code), and other relatively recent additions to C makes it EXTREMELY >difficult for many of us to actually USE that software. If there are long identifiers in use in smail or pathalias or some other key piece of software, I'd appreciate it if you'd point out where. I've compiled sendmail, pathalias, and smail on a UNIX PC running 3.0, which doesn't have flexnames, and it works fine. This suggests to me that there aren't any names used that aren't unique in the first 7 or 8 chars, as appropriate. If you have a counterexample, please let us know. If you have a compiler that can't handle 10 character names that ARE unique in the first 6 chars, it isn't a C compiler, it's a C subset. > The new mailers make heavy use of neat goodies like enumerated data >types, which many of us pcc based users can't deal with. Enums are in PCC and have been a part of the C language since V7 and UNIX 2.0. As far as I know, this means that enums are present in every version of UNIX in use today. (V6 and PWB don't count, they don't support UUCP without serious hacking.) There are restrictions on how enums can be used in some compilers, but smail stays within those restrictions. Again, if you're having problems, it would be helpful if you'd point out the specific piece of code causing the problem, and tell what version of UNIX on what hardware has trouble with it. I suppose it is possible that somebody has a braindamaged C compiler, since each UNIX port needs a new C compiler, but if you don't have enums, you don't conform to V7, System III, System V, Xenix, or 4BSD, all of which are supposed to have enums. Nonetheless, if the enums are causing a problem, and it's not just one locally braindamaged orphan system, let us know, it is possible to rewrite it to use #defines instead. In my experience, the only compilers out there today that don't support enums are those going by the K&R C book as the definition of the language, rather than the UNIX manual. As far as I know, all such compilers are for operating systems other than UNIX. I'd love to be corrected if I'm wrong. Mark