CLUDMP is the compiler, and is a normal ITS executable. It has a
variety of commands documented in CLU ORDER, but normally just the
file name can be passed on the JCL. The compiler emits an
intermediate CLUMAC file which is then assembled with MIDAS, leavning
a BIN file. The BIN file is not a normal ITS binary, but has to be
loaded into the CLU runtime.
The CLUSYS directory has files needed to assemble CLUMAC files. ALPHA
and OMEGA are inserted at the top and bottom, respectively. ALPHA in
turn needs PASS1, TYPES, and COMMON. It is not known how LOAD is
used, but it's also necessary in the compilation process.
TS CLUSYS is the runtime system. The procedure "fload" accepts a
string specifying a file to be loaded.
It's an open question whether the two executables can be rebuilt from
source code found on the scattered ITS backups.
Don't fail to compile H3TEXT if a NET corresponding to a HOST address
is missing, just whine. (As a comment in SYSHST;HOSTS3 says, the
network table is almost useless anyway.)
This version includes one new function to test graphs,
<MK&DSP>$ which compiles all the MCELLs in the SENE1 image and then displays
the complete SENE1 cell. It takes about 15 to 20 seconds to compile the
display. That's pretty amazing for source code, much faster then DM, or USC-ISI's pdp10s.
Just think what a difference the compiler will make...
With support in KLH10 (dpimp version 1.1.4 or later), read the IP configuration from the NOP sent. The IP address is in the network, source host, and source IMP fields, and the subnet mask size is in the "handling type" field. The first and last of these are very non-standard, but hey... it works.
This file was assembled from sources, and is identical to the KSHACK;
RAM 262 file. It has been tested on a real KS10. The older file
doesn't work on a real KS10.
DSD is the deselection device. It allows the PDP-6 and PDP-10 to
share some devices and "deselect" them when they are not in use.
CONI/O bit 0 (the sign bit) is set when a device is free to use.
This change ensures the bit 0 checks are only done when DSDP is
enabled.
Now that ITS will honor the DSDP switch, it might not check the CONI
DIS bit 0 to see if the 340 is available. This leads to a situation
where the display could be turned off but ITS thinks it's always
online and will try to use it.
This commit adds a check writing some CONO bits and read them back
with CONI.
This makes a small change in how the expect script interacts with PEEK
when it's built. Instead of triggering on text written by PEEK, it
waits for the $G that comes from the purify valret. The reason is
that if ITS considers the 340 online and available, PEEK running on
the console will send its output to the 340 rather than the teletype,
so expect has nothing to trigger on.
Instead, use ITSNMS table.
- MAGFRM doesn't need to check machine name.
- But still knows about MC's config...
- Look up hosts in ITSNMS table instead of a hardwired one, and use all of the ITSNMS for *.
Use bits in .RYEAR/.RLPDTM result to return a local time zone,
TZONE (definable in SYSTEM;CONFIG).
TZONE should be the integer timezone offset (hours West of UTC)
Bit 4.4 => timezone known (otherwise, assume EST/EDT = 5)
Bit 4.3 => sign bit of timezone offset
Bits 3.5-3.1 => absolute value of timezone offset
If DSTEU is defined and != 0, calculate daylight savings time
according to European Union rules:
- starts at 2:00 (standard time) on the last Sunday in March,
- ends at 2:00 (standard time) on the last Sunday in October.
In the default network config for klh10, replace up.update.uu.se with router.chaosnet.net, which is really the only one needed (since it knows everyone else, and tells you about them).
