[Info-vax] yet another sys$qiow question

[Jan-Erik Soderholm]

Den 2015-08-12 kl. 01:53, skrev Johnny Billquist:
> On 2015-08-12 00:38, kc at kayceesoftware.com wrote:
>> this is slightly related issue only as it pertains to the same serial
>> comm to a card reader.
>> to refresh: vms 8.4 IA64, tcpip 5.7 ECO 5, C 7.3-018
>> sys$qiow working fine now and seems very stable, my problem was getting
>> the terminator mask correct.
>>
>> I've been reading on sockets and will be converting to that mode soon for
>> another project and probably change this serial code at that point.
>>
>> BUT, in researching a very random NAK of my comm, i've found that the CRC
>> sometimes calcs a line feed (LF, ox0a), and the stored string shows ONLY
>> the LF i put in it....but the message that ends up at the card reader has
>> CRLF, which is then rejected as invalid.
>>
>> The terminal port (TNA1) is set pasthru, the VMS IO Ref Manual says
>> IO$M_NOFORMAT and TT2$M_PASTHRU are basically the same.
>>
>> So, is it the TT driver, or tcp, or something else converting the LF to
>> CRLF?
>
> While I cannot answer, I should repeat what I pointed out before. The
> "telnet" protocol is really just a TCP stream, with a couple of things
> thrown in.

You can use TNA ports/devices without using the telnet protocol.
Just use /prot=none.


> 0xff is special, and you are expected to use the network virtual
> terminal convention/protocol. One thing with that is about how to deal with
> CR and LF. And depending on various parameters and configurations, this
> could throw a wrench in your communication.
>
> Running tcpdump and capture the traffic would be the easiest way to confirm
> if the data transmitted contains a CR, or if the receiving end adds it.
>
> After that, you might want to continue hunting in greater depth wherever
> the trail leads you...
>
>      Johnny
>