• Talsiman/Postie

    From vorlon@21:1/195.1 to deon on Sunday, April 17, 2022 17:37:48
    Hi Deaon,

    Tresting Talsiman on linux/m68k and getting this error on the
    talsiman/postie end.
    2022-04-14 14:36:55 [1893] DEBUG: Starting Tosser
    2022-04-14 14:36:55 [1893] ERROR: Packet version != 2

    Hub is husky.
    1 14:36:15 Start tossing...
    A 14:36:15 Invalid pkt version 512!

    An endian issue?

    I was reading a gated newsgroup, that was talking about that in cpu's.

    The intresting thing is that hpt on the same machine works, but it can't
    read the squish files.

    Hpt on the same machine, can read & process JAM files though...

    I think Apam, didn't think someone would do what I've been playing around
    with though.



    \/orlon
    aka - Stephen




    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From deon@21:2/116 to vorlon on Sunday, April 17, 2022 20:28:12
    Re: Talsiman/Postie
    By: vorlon to deon on Sun Apr 17 2022 05:37 pm

    talsiman/postie end.
    2022-04-14 14:36:55 [1893] ERROR: Packet version != 2
    Hub is husky.
    A 14:36:15 Invalid pkt version 512!
    An endian issue?
    The intresting thing is that hpt on the same machine works, but it can't read the squish files.

    Hpt on the same machine, can read & process JAM files though...

    So that sounds even more like an endian issue.

    I'm assuming you compiled hpt and talisman on the m68k? I bet the hpt on the same machine sending something to a different machine wouldnt work... You'd get the same pkt version issue... (Send me a packet and we can see...)

    "512" is the 2nd bit of the 2nd byte, and "2" is the 2nd bit of the 1st byte, and in a fido packet, the "version" value is a 2 byte field.


    ...лоеп
    --- SBBSecho 3.15-Linux
    * Origin: I'm playing with ANSI+videotex - wanna play too? (21:2/116)
  • From vorlon@21:1/195.1 to deon on Monday, April 18, 2022 10:46:14
    Hi Deon,

    The intresting thing is that hpt on the same machine works, but
    it can't read the squish files.

    Hpt on the same machine, can read & process JAM files though...

    So that sounds even more like an endian issue.

    I'm assuming you compiled hpt and talisman on the m68k? I bet the
    hpt on the same machine sending something to a different machine
    wouldnt work... You'd get the same pkt version issue... (Send me a
    packet and we can see...)

    Yes, both got compiled on this machine that I'm on now. Linux Debian 11
    M68K. The hub is the fido hub, running husky on intel Debian 11 64bit.

    I thought of using hpt with Talisman, (hpt was already compiled) but even
    a fresh blank msg base hpt complained it could'nt read the squish mail
    files.






    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From deon@21:2/116 to vorlon on Monday, April 18, 2022 20:58:39
    Re: RE: Talsiman/Postie
    By: vorlon to deon on Mon Apr 18 2022 10:46 am

    Yes, both got compiled on this machine that I'm on now. Linux Debian 11 M68K. The hub is the fido hub, running husky on intel Debian 11 64bit.

    I thought of using hpt with Talisman, (hpt was already compiled) but even
    a fresh blank msg base hpt complained it could'nt read the squish mail files.

    hpt on m68k couldnt read "new" squish mail files (that it created)? Or you they were created (and populated) on the Intel and moved over?

    The later could also be endian related...


    ...лоеп
    --- SBBSecho 3.15-Linux
    * Origin: I'm playing with ANSI+videotex - wanna play too? (21:2/116)
  • From vorlon@21:1/195.1 to deon on Tuesday, April 19, 2022 14:50:50
    Re: RE: Talsiman/Postie
    By: vorlon to deon on Mon Apr 18 2022 10:46 am

    Yes, both got compiled on this machine that I'm on now. Linux
    Debian 11 M68K. The hub is the fido hub, running husky on intel

    Debian 11 64bit. I thought of using hpt with Talisman, (hpt was
    already compiled) but even a fresh blank msg base hpt complained
    it could'nt read the squish mail files.

    hpt on m68k couldnt read "new" squish mail files (that it created)?
    Or you they were created (and populated) on the Intel and moved
    over?

    I think things are getting confussed. Everything was compiled on the
    m68k, with only config files copied for a starting place.

    The message base files got created by the respective program.

    Hpt/Magicka = Work (Jam message base)
    Hpt/Talsiman = Not Work (Squish message base)
    Talsiman/Postie = Work, but the hub/postie reject the packets as bad.
    (Squish message base)

    Hpt can't read squish filse created by Talsiman/Postie.
    Talsiman/Postie can't read squish files created with hpt.

    Talisman dosn't support JAM message bases. That's how I found the Squish message base issue above, as hpt supports squish. But not in this configuration.






    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From deon@21:2/116 to vorlon on Tuesday, April 19, 2022 15:52:17
    Re: RE: Talsiman/Postie
    By: vorlon to deon on Tue Apr 19 2022 02:50 pm

    I think things are getting confussed. Everything was compiled on the
    m68k, with only config files copied for a starting place.

    Yeah I did get that. I just wanted to confirm that you didnt lift the squish (or jam) files from intel and was trying to read them from hpt compile on m68k.

    The message base files got created by the respective program.

    Got it.

    Hpt/Magicka = Work (Jam message base)
    Hpt/Talsiman = Not Work (Squish message base)
    Talsiman/Postie = Work, but the hub/postie reject the packets as bad. (Squish message base)

    Hpt can't read squish filse created by Talsiman/Postie.
    Talsiman/Postie can't read squish files created with hpt.

    Talisman dosn't support JAM message bases. That's how I found the Squish message base issue above, as hpt supports squish. But not in this configuration.

    I was trying to understand this:

    hpt(m68k) generates a packet and is read OK by hpt(intel) (or some other mail tosser I guess...)?

    can hpt(m68k) read it's own squish files OK?

    I think this will help to confirm whether endianness is an issue.

    I'm not sure what "Not Work" means (for Squish) - I dont recall if you described that error you were seeing. Not sure if there is a "debug mode" that might help..?

    Talisman/Postie rejecting the packet (as well as not reading the squish mail files) could still be an endian issue (the error you provided when opening a packet certainly fits that description), even if the answer to the first question is yes. Apam should jump in and comment, but it seems he's not around...


    ...лоеп
    --- SBBSecho 3.15-Linux
    * Origin: I'm playing with ANSI+videotex - wanna play too? (21:2/116)
  • From vorlon@21:1/195.1 to deon on Tuesday, April 19, 2022 22:46:31

    Hpt/Magicka = Work (Jam message base)
    Hpt/Talsiman = Not Work (Squish message base)
    Talsiman/Postie = Work, but the hub/postie reject the packets as bad.
    (Squish message base)

    Hpt can't read squish filse created by Talsiman/Postie.
    Talsiman/Postie can't read squish files created with hpt.

    Talisman dosn't support JAM message bases. That's how I found the
    Squish message base issue above, as hpt supports squish. But not in
    this configuration.

    I was trying to understand this:

    hpt(m68k) generates a packet and is read OK by hpt(intel) (or some
    other mail tosser I guess...)?

    Yes. The messages that came from Magicka/Hpt (This point) pass through the net633 ftn hub, that's running HPT (Intel).

    can hpt(m68k) read it's own squish files OK?

    Yes. This message is using the squish message base created by hpt. I compiled the msged editor to make sure it was doing that.

    I'm not sure what "Not Work" means (for Squish) - I dont recall if you described that error you were seeing. Not sure if there is a "debug
    mode" that might help..?

    Talisman can only read/write squish message bases.

    Talisman/Postie rejecting the packet (as well as not reading the
    squish mail files)

    Talisman/postie only can't read the squish mail files if they are created with hpt & vice versa.

    Talisman/Postie can read write them ok, but then the packet is rejected at the hub.

    could still be an endian issue (the error you provided when opening a packet certainly fits that
    description), even if the answer to the first question is yes.
    Apam should jump in and comment, but it seems he's not around...

    Until Apam show's up, we wont know.

    --- msged/lnx 6.3 2022-02-08
    * Origin: Vorlon Empire: Sector 550 (21:1/195.1@fsxnet)
  • From vorlon@21:1/195.1 to deon on Saturday, April 23, 2022 00:09:50
    HI Daon.

    Did you see the fidonet netmail I sent you with the two zip files?





    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From deon@21:2/116 to vorlon on Saturday, April 23, 2022 22:22:33
    Re: RE: Talsiman/Postie
    By: vorlon to deon on Sat Apr 23 2022 12:09 am

    HI Daon.

    Did you see the fidonet netmail I sent you with the two zip files?

    Howdy, I got it. Didnt look at the files yet.

    (I normally access the BBS with iTerm, that doesnt support zmodem - but I'm just firing up Syncterm now.)

    ...

    Yup, I reckon it is endian related:
    hpt (works, shows this):

    00000000 c3 00 c3 00 e6 07 03 00 12 00 0a 00 2f 00 0e 00 |o?=.o?=.o?=......./...|
    00000010 00 00 02 00 01 00 01 00 ff 01 00 00 00 00 00 00 |........o?=.......|
    00000020 00 00 15 00 15 00 00 00 00 01 10 09 01 00 15 00 |................| 00000030 15 00 01 00 00 00 00 00 00 00 02 00 c3 00 c3 00 |............o?=.o?=.|

    postie (didnt work) shows this:

    00000000 00 c3 00 c3 07 e6 00 03 00 0e 00 11 00 0e 00 31 |.o?=.o?=.o?=.........1|
    00000010 00 00 00 02 ff ff 00 01 fe 01 00 00 00 00 00 00 |....o?=o?=..o?=.......|
    00000020 00 00 00 15 00 15 00 01 01 00 00 00 00 01 00 15 |................| 00000030 00 15 00 01 00 00 45 54 53 50 00 02 00 c3 00 c3 |......ETSP...o?=.o?=|

    The first 4 bytes are the from and to address and they are reversed. The next few bytes are year, month, date and they too are reversed.

    Is there a compile option to change endianess?


    ...лоеп
    --- SBBSecho 3.15-Linux
    * Origin: I'm playing with ANSI+videotex - wanna play too? (21:2/116)
  • From apam@21:1/101 to vorlon on Friday, June 17, 2022 15:48:14
    Until Apam show's up, we wont know.

    Talisman / Postie doesn't do big endian. There's no byte swapping happening, which is why you're saw what you saw. I could add it I suppose, but I it's not really a priority.

    Andrew

    --- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
    * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)
  • From vorlon@21:1/195.1 to apam on Saturday, June 18, 2022 10:37:12
    Until Apam show's up, we wont know.

    Talisman / Postie doesn't do big endian. There's no byte swapping happening, which is why you're saw what you saw. I could add it I
    suppose, but I it's not really a priority.

    Let me first say I'm not a programmer...

    Looking through the Magicka source (I've moved all the data files to the
    data directory), I see that Magicka does do a byte-swap command. Is that
    the only thing that needs to happen or is it more in-depth?




    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From apam@21:1/101 to vorlon on Saturday, June 18, 2022 14:30:53
    Looking through the Magicka source (I've moved all the data files to the data directory), I see that Magicka does do a byte-swap command. Is that the only thing that needs to happen or is it more in-depth?

    Packets / squish message bases are in little endian format, so they work as is on little endian computers. So basically, for big endian, every time you read a multibyte number from a packet or message base you need to swap the bytes around to suit the cpu. So having a byte swap command is important, but you also need to find all the places that read / write data and have a check for cpu type and swap bytes.

    It's not hard to do, just time consuming, ideally I would have thought about it when writing the tosser / bbs and added it then.

    I can look at adding it in, but I don't have a big endian computer to test it on, so would likely need some help to determine if i've found all the places that need byte swapping.

    Andrew

    --- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
    * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)
  • From vorlon@21:1/195.1 to apam on Sunday, June 19, 2022 11:07:10
    It's not hard to do, just time consuming, ideally I would have
    thought about it when writing the tosser / bbs and added it then.

    That's ok. I bet you never thought that someone would spin up a system on
    such old hardware. #-)


    I can look at adding it in, but I don't have a big endian computer
    to test it on, so would likely need some help to determine if i've
    found all the places that need byte swapping.

    There's a couple of options.

    1: I give you ssh access. I have both real hardware and emulated setup's.

    Real: 68040@40Mhz, 132Mb of Ram
    Emu: 68040@1.2Ghz, 1Gb of Ram

    2: You spin up a emulated setup.

    https://wiki.debian.org/M68k/QemuSystemM68k

    Use this iso though, as it's the latest. Then it's just a matter of
    updating the install like any debian linux system.

    https://cdimage.debian.org/cdimage/ports/current/debian-11.0.0-m68k-NETINS T-1.iso









    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)
  • From apam@21:1/101 to vorlon on Sunday, June 19, 2022 18:53:11
    That's ok. I bet you never thought that someone would spin up a system on such old hardware. #-)

    Nah, I think I was just focused on making it portable to windows this time, and didn't think about anything other than windows and linux.

    I've spun up a qemu-m68k vm here, and done some work in a branch called "bigendian"

    There should be enough to get it going with husky now, but I haven't tested that.

    There's a lot more to do, for example, QWK and Bluewave probably wont work, QWK networking, WWIV networking and Postie probably wont work, binki might, but i'm not sure on that.

    the squish code is shared amoungst the different programs so that side of thing should work, it mostly just interfacing with various packet formats now.

    if you want to try it out / follow what i'm doing just issue 'git checkout bigendian' after doing a clone and that will set you on the branch.

    Andrew

    --- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
    * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)
  • From apam@21:1/101 to apam on Sunday, June 19, 2022 19:43:36
    There's a lot more to do, for example, QWK and Bluewave probably wont

    Actually QWK / QWK net should work I think, as it's all stored in bytes. bluewave i've just done a fix that should make it work properly (the endian stuff was already there, but I think the detection was off)

    Andrew

    --- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
    * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)
  • From apam@21:1/101 to vorlon on Sunday, June 19, 2022 21:59:50
    Everything should now be working in the bigendian branch.

    I'll merge it soon. You should be good to go for m68k now.

    Andrew

    --- Mystic BBS v1.12 A47 2021/12/24 (Linux/64)
    * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)
  • From vorlon@21:1/195.1 to apam on Tuesday, June 21, 2022 10:02:08
    That's ok. I bet you never thought that someone would spin up
    a system on such old hardware. #-)

    Nah, I think I was just focused on making it portable to windows
    this time, and didn't think about anything other than windows and
    linux.


    Well the Amiga 3000 is running Linux, just not not on intel/amd hardware.


    I've spun up a qemu-m68k vm here, and done some work in a branch
    called "bigendian"

    [...]


    if you want to try it out / follow what i'm doing just issue 'git
    checkout bigendian' after doing a clone and that will set you on
    the branch.

    Wow! I guess we know what you got upto on sunday/monday. I see you've now released .40.. I'll grab that and do a compile, and test...




    \/orlon



    --- MagickaBBS v0.15alpha (Linux/m68k)
    * Origin: Vorlon Empire (21:1/195.1)