home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!cel.cummins.com!philip
- From: philip@cel.cummins.com (Philip D. Pokorny)
- Newsgroups: comp.sys.apollo
- Subject: Re: rz/sz doesn't work on our machiens
- Message-ID: <9209082141.AA02720@cel.cummins.com>
- Date: 8 Sep 92 21:41:28 GMT
- Sender: daemon@ucbvax.BERKELEY.EDU
- Distribution: world
- Organization: The Internet
- Lines: 47
-
- Chuck Tomasi asks:
- >
- > We're running mostly DN3550/DN4500/DN5500 machines on our site (with a
- > few 425Ts) under Domain/OS 10.3.5. We compiled the rz/sz sources with
- > SYSTYPE=sys5.3 and everything appeared to go alright. The problem is
- You're doing better than me... I didn't try very hard but
- didn't get very far either...
-
- > that they don't work. The same sources work fine from an HP-UX v8.0
- > machine however it seems that something is being blocked or trapped when
- > we invoke them from the Domain machines.
- >
- > Let me explain. We are working on a Mac from home with software which
- > supports auto-zmodem downloading. From the HP-UX machine the zmodem
- > init string initiates the zmodem transfer and everything goes as
- > expected, however with the same command is issued from the Domain
- > machine you see the same portion of the init string, but you also see
- > things like the file name, CRC, and a few other things. This is not
- > normally seen!
- You don't say whether the transfer commences or not... Are the
- files transfered correctly?
-
- I don't normally like to say negative things about companies,
- but... We bought a product called YAM (a communications program)
- from a company called Omen Technologies. This is the same
- company that developed and wrote the ZMODEM protocols. (Check
- the headers) In porting the YAM software to the Apollo, I was
- witness to some of the worst coding practices ever performed in
- C. (This from a company that bills itself as the "High
- Reliability Software Company") In any case, the protocols
- didn't work. There is a routine called readline that reads a
- line of input and returns after reading a line or a timeout is
- exceeded. This turned out to be the culprit. Something about
- signal()'s not acting correctly and needing both BLOCK and
- NOBLOCK access to an sio. I "fixed" it by writing a BUSY loop
- that checks time_$clock and does NON-BLOCKING reads. It's a
- horrible waste of CPU cycles but if works on the 425t's of the
- only users of the software. You might check the readline
- function (or equivalent read with timeout function) to see if
- that is your problem. If you come up with a better fix than the
- BUSY loop I'm using, I'd appreciate hearing from you so I can
- "fix it right".
- Sincerely,
- Philip D. Pokorny
- philip@cel.cummins.com
- :)
-
-