How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on simh I can boot into. I know how to mount a tape and use the BACKUP program to restore it. I can't figure out how to insert files from my Ubuntu into a .tap file, so I can load it on my TOPS-10. The files are the source for "MUD1" and "Empire" both of which can be compiled on TOPS-10. I know how to compile MUD1, I have a guide for it. Just need to put the source files (from Github) into a .tap. I know that the Github repo has .tap files, but those have typos and errors in the source .MAS files.

By the way if anyone knows how to build Empire from source, please tell me what commands to type. It's in FORTRAN-10.

On Wednesday, September 13, 2023 at 12:03:39 PM UTC-7, Kroma Zomal wrote:

> By the way if anyone knows how to build Empire from source, please tell me what commands to type. It's in FORTRAN-10.

I know at least one person who knows, and I thought he followed this, but maybe not.

Walter Bright is the one who wrote it.

On Wednesday, September 13, 2023 at 12:03:39 PM UTC-7, Kroma Zomal wrote:
> How can I load files onto a tape for TOPS-10?

Sounds like the problem I had 45 years ago.

I used to write tapes on IBM S/370 systems and bring them to TOPS-10.

The first problem I had was getting an 800 BPI tape. That is related to
a funny way that OS/360 works.

The second was reading 8 bit EBCDIC on TOPS-10.
I had a Fortran program to read tapes, and convert them.

There are a few different ways to read/write tapes on TOPS-10,
all of which I knew 45 years ago.

Kroma Zomal <kromazomal@gmail.com> writes:
>How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on s=
>imh I can boot into. I know how to mount a tape and use the BACKUP program =
>to restore it. I can't figure out how to insert files from my Ubuntu into a=
> .tap file, so I can load it on my TOPS-10. The files are the source for "M=
>UD1" and "Empire" both of which can be compiled on TOPS-10. I know how to c=
>ompile MUD1, I have a guide for it. Just need to put the source files (from=
> Github) into a .tap. I know that the Github repo has .tap files, but those=
> have typos and errors in the source .MAS files.
>
>By the way if anyone knows how to build Empire from source, please tell me =
>what commands to type. It's in FORTRAN-10.

There are two aspects to your question:

1) What is the format of a SIMH tape file on the SIMH host?

http://simh.trailing-edge.com/docs/simh_magtape.pdf

2) What is the format of the data on a TOPS-10 tape? It's probably
not the traditional ANSI format.

https://www.pcjs.org/software/dec/pdp10/tapes/
https://webstore.ansi.org/standards/incits/ansix32241994

Kroma Zomal <kromazomal@gmail.com> writes:

> How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on
> simh I can boot into. I know how to mount a tape and use the BACKUP program
> to restore it. I can't figure out how to insert files from my Ubuntu into a
> .tap file, so I can load it on my TOPS-10. The files are the source for
> "MUD1" and "Empire" both of which can be compiled on TOPS-10. I know how to
> compile MUD1, I have a guide for it. Just need to put the source files (from
> Github) into a .tap. I know that the Github repo has .tap files, but those
> have typos and errors in the source .MAS files.

The best way is to treat each Linux file separately, rather than getting into
the arcane details of building a BACKUP format tape. Trust me on this.

A SimH tape file is a stream of bytes starting with a 4 byte little endian
length of a block, followed by the native format data, followed by a repeat of
the 4 byte length.

The data on a Tops-10 (or TOPS-20) tape can be represented several different
ways, but the most common is called "coredump" format, and presents the 36 bits
of data in a word as five 8 bit frames as follows

b0 b1 b2 b3 b4 b5 b6 b7
b8 b9 b10 b11 b12 b13 b14 b15
b16 b17 b18 b19 b20 b21 b22 b23
b24 b25 b26 b27 b28 b29 b30 b31
0 0 0 0 b32 b33 b34 b35

Since you are presumably dealing with ASCII data, you will need to turn your
Linx 8 bit characters into PDP-10 7 bit characters before packing them into a
36 bit word to chop them up into a set of tape frames.

Or you could use Kermit and avoid tape images altogether.

> By the way if anyone knows how to build Empire from source, please tell me
> what commands to type. It's in FORTRAN-10.

.compile e1.for,e2.for,e3.for <etc.>
.load

or .load e1.for,e2.for <etc.>

where the load command first compiles the source files which are newer than the
object files which may be present, then calls LINK on all the object files (.REL).
(In the first case, the system remembers the command line from the compile step
and feeds it to the load command.)

It's been about 10 years since I looked at EMPIRE, so the above is just a sketch.

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

On Wednesday, September 13, 2023 at 1:30:30 PM UTC-7, Rich Alderson wrote:

(snip)

> The data on a Tops-10 (or TOPS-20) tape can be represented several different
> ways, but the most common is called "coredump" format, and presents the 36 bits
> of data in a word as five 8 bit frames as follows
>
> b0 b1 b2 b3 b4 b5 b6 b7
> b8 b9 b10 b11 b12 b13 b14 b15
> b16 b17 b18 b19 b20 b21 b22 b23
> b24 b25 b26 b27 b28 b29 b30 b31
> 0 0 0 0 b32 b33 b34 b35
Is this the form you get with the COPY command?

I remember sometimes wanting to test tapes that I didn't know about,
so copying the COBOL compiler onto the tape, and then back off again.
(Or maybe it was the library.) It was the biggest file on the system I
was using available for copying.

But for reading actual EBCDIC tapes, there is a way to read four tape
bytes into a 36 bit word. I would do that in Fortran, and then extract
the bytes, and do EBCDIC to ASCII conversion.

A favorite for OS/360 is FB80, where the tape has fixed length record
of 80 characters, blocked into some multiple of 80 bytes. That is
the usual text format for OS/360, and the form that most compiler
want as input. And so I had programs to read that.

gah4 <gah4@u.washington.edu> writes:

> On Wednesday, September 13, 2023 at 1:30:30=E2=80=AFPM UTC-7, Rich Alderson wrote:
>
> (snip)
>
>> The data on a Tops-10 (or TOPS-20) tape can be represented several different
>> ways, but the most common is called "coredump" format, and presents the 36 bits
>> of data in a word as five 8 bit frames as follows

>> b0 b1 b2 b3 b4 b5 b6 b7
>> b8 b9 b10 b11 b12 b13 b14 b15
>> b16 b17 b18 b19 b20 b21 b22 b23
>> b24 b25 b26 b27 b28 b29 b30 b31
>> 0 0 0 0 b32 b33 b34 b35

> Is this the form you get with the COPY command?

Yes.

> I remember sometimes wanting to test tapes that I didn't know about,
> so copying the COBOL compiler onto the tape, and then back off again.
> (Or maybe it was the library.) It was the biggest file on the system I
> was using available for copying.

> But for reading actual EBCDIC tapes, there is a way to read four tape
> bytes into a 36 bit word. I would do that in Fortran, and then extract
> the bytes, and do EBCDIC to ASCII conversion.

> A favorite for OS/360 is FB80, where the tape has fixed length record
> of 80 characters, blocked into some multiple of 80 bytes. That is
> the usual text format for OS/360, and the form that most compiler
> want as input. And so I had programs to read that.

The issue with the "industry standard" format (as it is called in the DEC
manuals) is that a program will be necessary to convert the 8 bit ASCII
characters to 7 bits and pack them 5 to a word, which is at least as difficult
as using coredump format.

I still think that Kermit is the proper way to do this.

--
Rich Alderson news@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

On Wednesday, September 13, 2023 at 1:30:30 PM UTC-7, Rich Alderson wrote:

(snip)

> The data on a Tops-10 (or TOPS-20) tape can be represented several different
> ways, but the most common is called "coredump" format, and presents the 36 bits
> of data in a word as five 8 bit frames as follows

> b0 b1 b2 b3 b4 b5 b6 b7
> b8 b9 b10 b11 b12 b13 b14 b15
> b16 b17 b18 b19 b20 b21 b22 b23
> b24 b25 b26 b27 b28 b29 b30 b31
> 0 0 0 0 b32 b33 b34 b35

OK, having not thought about this for many years, are lines padded to whole words?

I do remember that SOS puts a line number as a five digit number in one word,
with the low bit set so programs will know. That seems to suggest that it has
to fill up the previous word first.

> Since you are presumably dealing with ASCII data, you will need to turn your
> Linx 8 bit characters into PDP-10 7 bit characters before packing them into a
> 36 bit word to chop them up into a set of tape frames.
> Or you could use Kermit and avoid tape images altogether.

If you have an emulated serial port, that does seem a good way.

Rich Alderson wrote:
> Kroma Zomal wrote:
>> How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on
>> simh I can boot into. I know how to mount a tape and use the BACKUP program
>
> The best way is to treat each Linux file separately, rather than getting into
> the arcane details of building a BACKUP format tape. Trust me on this.

That's a pretty good suggestion. I have occasionally made BACKUP tapes
to move files to TOPS-10, but it's always a pain because I keep
forgetting which tool to use, and how to operate it.

However, now we need to explain how to implement your suggestion. How
to convert each text file to core dump format, getting them on a tape
(or tapes) image, and then something like COPY MTA0: E1.FOR for each
file.

I *think* you can do the ASCII to core dump conversion with
cat10 -Wascii -Xcore < e1.for > e1.core
using https://github.com/larsbrinkhoff/pdp10-its-disassembler

And then using https://github.com/brouhaha/tapeutils:
tapewrite -s output.tap e1.core e2.core ...

Is some particular tape block size needed?

gah4 wrote:
> OK, having not thought about this for many years, are lines padded to
> whole words?
>
> I do remember that SOS puts a line number as a five digit number in
> one word, with the low bit set so programs will know. That seems to
> suggest that it has to fill up the previous word first.

For that kind of text file, yes, lines are padded out with NUL
characters.

For text files without SOS line numbers, no padding.

WAITS has E text files with padding between pages.

On Wednesday, September 13, 2023 at 11:16:51 PM UTC-7, Lars Brinkhoff wrote:
> gah4 wrote:
> > OK, having not thought about this for many years, are lines padded to
> > whole words?
> > I do remember that SOS puts a line number as a five digit number in
> > one word, with the low bit set so programs will know. That seems to
> > suggest that it has to fill up the previous word first.

> For that kind of text file, yes, lines are padded out with NUL
> characters.
> For text files without SOS line numbers, no padding.

OK, what I might have known 45 years ago.

Do files without line numbers, but with padding, still work?

I never used TECO for TOPS-10, but did when I got to using RT-11.

> WAITS has E text files with padding between pages.

gah4 wrote:
> Do files without line numbers, but with padding, still work?

Sorry, I don't know really. I would imagine that TOPS-10 and 20 editors
ignore the NUL's. I'm more of an ITS user, and over here text file
characters are processed and diplayed verbatim, except TECO's Y command
(without the @ modifier) treats ^L as a page separator.

On Wednesday, September 13, 2023 at 4:30:30 PM UTC-4, Rich Alderson wrote:
> Kroma Zomal <kroma...@gmail.com> writes:
>
> > How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on
> > simh I can boot into. I know how to mount a tape and use the BACKUP program
> > to restore it. I can't figure out how to insert files from my Ubuntu into a
> > .tap file, so I can load it on my TOPS-10. The files are the source for
> > "MUD1" and "Empire" both of which can be compiled on TOPS-10. I know how to
> > compile MUD1, I have a guide for it. Just need to put the source files (from
> > Github) into a .tap. I know that the Github repo has .tap files, but those
> > have typos and errors in the source .MAS files.
> The best way is to treat each Linux file separately, rather than getting into
> the arcane details of building a BACKUP format tape. Trust me on this.
>
> A SimH tape file is a stream of bytes starting with a 4 byte little endian
> length of a block, followed by the native format data, followed by a repeat of
> the 4 byte length.
>
> The data on a Tops-10 (or TOPS-20) tape can be represented several different
> ways, but the most common is called "coredump" format, and presents the 36 bits
> of data in a word as five 8 bit frames as follows
>
> b0 b1 b2 b3 b4 b5 b6 b7
> b8 b9 b10 b11 b12 b13 b14 b15
> b16 b17 b18 b19 b20 b21 b22 b23
> b24 b25 b26 b27 b28 b29 b30 b31
> 0 0 0 0 b32 b33 b34 b35
>
> Since you are presumably dealing with ASCII data, you will need to turn your
> Linx 8 bit characters into PDP-10 7 bit characters before packing them into a
> 36 bit word to chop them up into a set of tape frames.
>
> Or you could use Kermit and avoid tape images altogether.
> > By the way if anyone knows how to build Empire from source, please tell me
> > what commands to type. It's in FORTRAN-10.
> .compile e1.for,e2.for,e3.for <etc.>
> .load
>
> or .load e1.for,e2.for <etc.>
>
> where the load command first compiles the source files which are newer than the
> object files which may be present, then calls LINK on all the object files (.REL).
> (In the first case, the system remembers the command line from the compile step
> and feeds it to the load command.)
>
> It's been about 10 years since I looked at EMPIRE, so the above is just a sketch.
>
> --
> Rich Alderson ne...@alderson.users.panix.com
> Audendum est, et veritas investiganda; quam etiamsi non assequamur,
> omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
> --Galen

Seriously consider the Kermit suggestion by Rich if you want to focus on your main project and not get wrapped up in the side project.

For the uninitiated, Kermit has a "server" mode. From the source computer I'd Kermit into the destination machine and login. Next on the destination side, run Kermit and say "server". Next, escape-character back to the source machine, and SEND *.FOR or whatever (I didn't have to type each filename), and let Kermit deal with the ASCII conversions and such. You can also GET *.FOR. There is a good explanation of these operations in the user guide for the '20 at http://www.columbia.edu/kermit/ftp/d/k20mit.pdf, although most of it pretty much transfers to any OS.

On 2023-09-14 14:54, r cs wrote:
> On Wednesday, September 13, 2023 at 4:30:30 PM UTC-4, Rich Alderson wrote:
>> Kroma Zomal <kroma...@gmail.com> writes:
>>
>>> How can I load files onto a tape for TOPS-10? I have a working TOPS-10 on
>>> simh I can boot into. I know how to mount a tape and use the BACKUP program
>>> to restore it. I can't figure out how to insert files from my Ubuntu into a
>>> .tap file, so I can load it on my TOPS-10. The files are the source for
>>> "MUD1" and "Empire" both of which can be compiled on TOPS-10. I know how to
>>> compile MUD1, I have a guide for it. Just need to put the source files (from
>>> Github) into a .tap. I know that the Github repo has .tap files, but those
>>> have typos and errors in the source .MAS files.
>> The best way is to treat each Linux file separately, rather than getting into
>> the arcane details of building a BACKUP format tape. Trust me on this.
>>
>> A SimH tape file is a stream of bytes starting with a 4 byte little endian
>> length of a block, followed by the native format data, followed by a repeat of
>> the 4 byte length.
>>
>> The data on a Tops-10 (or TOPS-20) tape can be represented several different
>> ways, but the most common is called "coredump" format, and presents the 36 bits
>> of data in a word as five 8 bit frames as follows
>>
>> b0 b1 b2 b3 b4 b5 b6 b7
>> b8 b9 b10 b11 b12 b13 b14 b15
>> b16 b17 b18 b19 b20 b21 b22 b23
>> b24 b25 b26 b27 b28 b29 b30 b31
>> 0 0 0 0 b32 b33 b34 b35
>>
>> Since you are presumably dealing with ASCII data, you will need to turn your
>> Linx 8 bit characters into PDP-10 7 bit characters before packing them into a
>> 36 bit word to chop them up into a set of tape frames.
>>
>> Or you could use Kermit and avoid tape images altogether.
>>> By the way if anyone knows how to build Empire from source, please tell me
>>> what commands to type. It's in FORTRAN-10.
>> .compile e1.for,e2.for,e3.for <etc.>
>> .load
>>
>> or .load e1.for,e2.for <etc.>
>>
>> where the load command first compiles the source files which are newer than the
>> object files which may be present, then calls LINK on all the object files (.REL).
>> (In the first case, the system remembers the command line from the compile step
>> and feeds it to the load command.)
>>
>> It's been about 10 years since I looked at EMPIRE, so the above is just a sketch.
>>
>> --
>> Rich Alderson ne...@alderson.users.panix.com
>> Audendum est, et veritas investiganda; quam etiamsi non assequamur,
>> omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
>> --Galen
>
> Seriously consider the Kermit suggestion by Rich if you want to focus on your main project and not get wrapped up in the side project.
>
> For the uninitiated, Kermit has a "server" mode. From the source computer I'd Kermit into the destination machine and login. Next on the destination side, run Kermit and say "server". Next, escape-character back to the source machine, and SEND *.FOR or whatever (I didn't have to type each filename), and let Kermit deal with the ASCII conversions and such. You can also GET *.FOR. There is a good explanation of these operations in the user guide for the '20 at http://www.columbia.edu/kermit/ftp/d/k20mit.pdf, although most of it pretty much transfers to any OS.

I can only agree. Rich suggestion of using Kermit is what I'd go for.
Going via tape is just making things very complicated.

Johnny

The simtools repository has a "backup" program to read TOPS-10 tapes, which I know works, but sub directories can be tricky.

https://github.com/simh/simtools

I have also used a program called "backwr" to read TOPS-10 tapes. I believe it came from Arthur Krewat's site:

http://pdp10.kilonet.org/

Both programs needed a bit of experimentation, but I think they worked last time I tried.

Kermit also works well, if slow.

David.

Update: Kermit worked! I didn't have it on my TOPS-10 but luckily found a .tap here https://sky-visions.com/dec/tops704/tapes/k10mit-135.tap
Kermit itself was quite slow but I'm pleased at how easy it was to use
Only thing now is to compile Empire and MUD1. Unfortunately I'm having problems with both, but I've contacted those who uploaded them to Github since they most likely know how to do it correctly

On Friday, September 15, 2023 at 4:37:34 AM UTC-4, Kroma Zomal wrote:
> Update: Kermit worked! I didn't have it on my TOPS-10 but luckily found a .tap here https://sky-visions.com/dec/tops704/tapes/k10mit-135.tap
> Kermit itself was quite slow but I'm pleased at how easy it was to use
> Only thing now is to compile Empire and MUD1. Unfortunately I'm having problems with both, but I've contacted those who uploaded them to Github since they most likely know how to do it correctly

Congratulations! Thanks for passing along the .tap link. Let us know about how you resolve the Empire and MUD1 builds.

Regards,
rcs

Finally managed to build MUD1, from the '86 sources in the master branch on Github, which were transferred via Kermit

This is probably not the standard/correct way to do it but this is what worked for me. Previously I faced errors with the missing MUD..PM file, elevated rights, or simply could not load into MUD.EXE

Note you must have BCPL installed and configured beforehand which is covered on Quentin's blog https://www.quentin.org.uk/ ,check out the retro computing posts, I suggest reading all of them, there is info in the TOPS-10 FAQ and comments too

-git clone https://github.com/PDP-10/MUD1
-delete the .git and .gitignore files (or just DL the zip instead)
-Kermit over the files to TOPS-10, place them in [2011,2776]
-make a new user 'richard' with PPN [2011,2776] in REACT, give him ENQ/DEQ 500, IPCF 500 500 500
-login with operator, cd over to [2011,2776] with setsrc
-do mud all to build MUD
-Kermit messes up the .txt files and .get files, open them all one by one with SOS and re-save them (command is ecb)
-run dbase to build the database
-run mud as operator, if that doesn't work, login as richard and run mud, if it still doesn't work just restart simh and try run mud with richard again. Worked for me after restarting.
-experience gaming history

On Wednesday, September 13, 2023 at 11:16:51 PM UTC-7, Lars Brinkhoff wrote:
> gah4 wrote:
> > OK, having not thought about this for many years, are lines padded to
> > whole words?
> > I do remember that SOS puts a line number as a five digit number in
> > one word, with the low bit set so programs will know. That seems to
> > suggest that it has to fill up the previous word first.

> For that kind of text file, yes, lines are padded out with NUL
> characters.
> For text files without SOS line numbers, no padding.
> WAITS has E text files with padding between pages.

I was just reading the Fortran-10 manual, unrelated to this question, and noticed
that it does say that lines will be null padded to whole words. It seems that is the
way it writes files, and as far as I can tell, the way it expects to read them.