Kermit az Itanium VMS-en

Hátha más is emléxik még :-)


fodor31@freemail.hu Mappa: kuka 14. levél 

Dátum: Thu, 8 Apr 2004 11:22:17 -0400
Feladó: "Skonetski, Susan" <susan.skonetski@hp.com>
Címzett: "Skonetski, Susan" <susan.skonetski@hp.com>
Tárgy: OpenVMS Pearl - Kermit runs on OpenVMS I64 on HP Integrity 
(Itanium) Servers
 

 

Many thanks to John Egolf (HP OpenVMS Team) and Frank da Cruz 
OpenVMS Champion.
 
Warm Regards,
Sue
 
-----Original Message-----
From: Egolf, John 
Sent: Thursday, April 08, 2004 9:44 AM
To: Skonetski, Susan
Subject: FW: Pass II - RE: Kermit runs on OpenVMS I64 on HP Integrity 
(Itanium) Servers


Sue,
 
Frank da Cruz has given his permission for this to go out to the 'world'.
 
/John



---------------------------------------------------------------------------
-----


The Kermit Project at Columbia University, which first created the Kermit
 
file transfer protocol in 1981, along with software to implement it, has 
been developing the protocol and software ever since, until now it can 
be found in some form on almost every computer in the world, not to 
mention other devices such as cash registers, milling machines, and 
heart monitors.

Among Columbia's premiere Kermit implementations is its portable C-
language version, C-Kermit for Unix and OpenVMS, currently running on 
about 500 different hardware-OS combinations, supporting network as 
well as serial communications and including a built-in programming 
language to allow full automation of any communications task:

http://www.columbia.edu/kermit/ckermit.html

In preparation for the next C-Kermit release, Frank da Cruz, manager 
and founder of the Kermit Project and principal author of C-Kermit, had 
a brand-new platform to accommodate: OpenVMS I64 8.1 on HP 
Integrity (Itanium) Servers.

When asked about his experiences regarding the port, Frank says 

 

"C-Kermit has about 300,000 lines of code and is intended to be 
portable to all VMS and Unix versions, as well as to several other 
operating system families now mostly defunct.

 

In other words, it was pretty portable to begin with.  In particular 
we've been building and running on VMS/Alpha ever since the Alpha 
first appeared, so by now we're far beyond VAXC/DECC and 32/64-bit 
issues.

 

All I did was bring the sources to HP Test Drive system and run the 
build procedure.

 

It blew up in the platform-dependent modules because it had 
previously coded to test for VAX or Alpha and to give up if neither, so all
 
such tests had to be expanded to include IA64.  I had to hunt around 
in the header files to find that the appropriate compile-time symbol was 
__ia64.

 

Then I added code to the (DCL) build procedure to detect VMS 8.x and 
pass this along to the code (currently used only for runtime 
diagnostics).

 

At this point it built OK and ran OK, at least in limited testing: it start
s 
normally, can make Telnet connections, and can transfer files.

 

The only glitch I found was in reporting file-transfer statistics.  The 
elapsed time was off the map.  To get this information I had been using 
sys$gettim(), lib$sub_times(), and lib$cvtf_from_internal_time(), which 
appear to be returning bogus values.  I reported this to OpenVMS 
engineering, and worked around it by using the Unix gettimeofday() 
API, which seems to work fine in VMS 8.1.

 

All done in two days!  Working sample -- source code and binaries are 
available."


 

Frank took advantage of HP's Test Drive system, which enables 
OpenVMS customers / partners to get hands-on 'play-time' to see what 
it is like to port code onto the newest version of OpenVMS on the new 
Intel Itanium2 hardware architecture.  Frank's positive experiences are 
shared by others porting their applications. 

For more information on the Kermit Project and software, visit: 
http://www.columbia.edu/kermit/kermit.html