I've spent a considerable amount of time porting various Unix C software to Xenix 286. My main task has been to identify and correct the implicit assumptions that pointers and integers are 32-bit, as opposed to 16-bit, and to ensure that there are no integer overflows. Some software was well-written, utilizing portable types and macros, while other software was a mess.
It was for work. I worked for a small ISP and they tried to use 286 machines as terminals and mail servers. So I ported stuff like CNEWS, UUCP and bunch of related stuff. It was around 1994.
> the ability to change the switch operator from / to - like XENIX/UNIX, and the ability to change the path separator from \ to /
I first learned about this from an article in the March/April 1984 issue of the newsletter published by Carousel Microtools (a company that implemented Kernighan/Plauger Software Tools from the book of the same name).
"Fortunately there is a way to suppress this confusion
altogether. PC-DOS and MS-DOS version 2.x share a feature
that is little known and which IBM never mentions."
"By the end of 1984, Tandy was the largest UNIX systems vendor of their day"
My high school had I think the TRS-80 Model 16B with I thought 8-12 terminals attached. It was primarily used for a COBOL class, replacing a Burroughs mini and punch cards.
I never personally used Xenix, but I did use SCO openserver & unixware, back in the day working for a billing company.
We had a bank of modems wired up, and the system would maintain SSH tunnels to remote sites, then use UUCP to transfer nightly batch-reports to a centralized location for processing.
My memory is hazy, but I remember having a hard time getting the system running on new hardware, and kernel updates were always a scary thing to apply - no change there I guess.
Even at the time though Linux (Debian) was starting to replace these machines due to cost grounds.
My first sysadmin job ( along with webmaster and HTML/CGI programmer introduced me to UNIX and Xenix in particular.
I immediately started spelunking around the system with my root account.
I soon discovered that something seemed odd with the use of system resources.
Poking some more I discovered some “hidden” folders.
Turns out someone hacked the ftp server and setup a warez and porn distribution.
(Possibly it could have been the former sysadmin who had set it up).
I notified higher ups, removed the content and removed the ftp server.
A few days later the server crashed hard.
Opened it up.
The hard disk was leaking fluid.
Notified the higher ups.
Found myself in a taxi hauling the server of considerable size and weight to a taxi.
We went to a contracted Xenix support center.
Tney said there was nothing they could do to recover files.
Stuffed in a replacement drive, installed a fresh version of the OS.
Said they would send the broken drive to a different company.
Hauling it back, get it setup.
Two days later, the ftp site was up and running again.
I repeated what I did the last time.
Then I sat down to read and understand the system at a far deeper level than I had so far.
“Despite this reduction in scope for MS-DOS 2.0, it did carry many bits of XENIX. The system adopted I/O redirection via less-than and greater-than symbols, piping, a hierarchical directory tree, file handles […]”
The source code for MSDOS 2 is available and the file descriptor stuff appears to be in https://github.com/microsoft/MS-DOS/blob/master/v2.0/source/... and XENIX2.ASM. It stands in contrast to the File Control Block API which MSDOS 1 (née 86-DOS) modeled after CP/M’s API.
11 comments
[ 3.9 ms ] story [ 36.1 ms ] threadI first learned about this from an article in the March/April 1984 issue of the newsletter published by Carousel Microtools (a company that implemented Kernighan/Plauger Software Tools from the book of the same name).
"Fortunately there is a way to suppress this confusion altogether. PC-DOS and MS-DOS version 2.x share a feature that is little known and which IBM never mentions."
My high school had I think the TRS-80 Model 16B with I thought 8-12 terminals attached. It was primarily used for a COBOL class, replacing a Burroughs mini and punch cards.
We had a bank of modems wired up, and the system would maintain SSH tunnels to remote sites, then use UUCP to transfer nightly batch-reports to a centralized location for processing.
My memory is hazy, but I remember having a hard time getting the system running on new hardware, and kernel updates were always a scary thing to apply - no change there I guess.
Even at the time though Linux (Debian) was starting to replace these machines due to cost grounds.
A few days later the server crashed hard. Opened it up. The hard disk was leaking fluid. Notified the higher ups. Found myself in a taxi hauling the server of considerable size and weight to a taxi. We went to a contracted Xenix support center. Tney said there was nothing they could do to recover files. Stuffed in a replacement drive, installed a fresh version of the OS. Said they would send the broken drive to a different company. Hauling it back, get it setup. Two days later, the ftp site was up and running again.
I repeated what I did the last time. Then I sat down to read and understand the system at a far deeper level than I had so far.
The source code for MSDOS 2 is available and the file descriptor stuff appears to be in https://github.com/microsoft/MS-DOS/blob/master/v2.0/source/... and XENIX2.ASM. It stands in contrast to the File Control Block API which MSDOS 1 (née 86-DOS) modeled after CP/M’s API.