Greetings All!

I had to move my point setup, I moved it to a rpi running manjaro.. all hpt stuff compiled no problem, had that areafix lib problem which I corrected, now however when tossing mail I get the following errors

SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!

So I thought maybe corrupted msgbase so I copied everything out of the dir so that I was starting fresh... However I still get the message above not for every area but some?

Thoughts?


----
Regards,


Rick Smith (Nitro)

... Now THAT's a dilemma! BBSing. Chocolate. BBSing. Chocolate.
--- GoldED+/LNX 1.1.5-b20180707
* Origin: ----> Abacus Sysop Point --->>>>bbs.abon.us:2323 (1:105/10.1)

Hello Rick,

Thursday April 08 2021, Rick Smith wrote to All:

I had to move my point setup, I moved it to a rpi running manjaro..
all hpt stuff compiled no problem, had that areafix lib problem which
I corrected, now however when tossing mail I get the following errors

SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!

So I thought maybe corrupted msgbase so I copied everything out of the
dir so that I was starting fresh...

Do you mean you started from the empty message base?

However I still get the message above not for every area but some? Thoughts?

If you can reproduce the error in one specific area with the empty message base, then please send me your full config without passwords to my email in the tagline.

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20180707
* Origin: Moscow, Russia (2:5020/1042)

Greetings Michael!

Friday April 09 2021 09:48, you wrote to me about an urgent matter!:

* Forwarded from area 'fidosoft.husky'
Hello Rick,

SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!
A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!

Do you mean you started from the empty message base?

Ok sorry about that it wasnt clear, I installed hpt and bink fresh and then copied over the the configs and msgbase etc from a working intall to the new install. So initially when I got the smapi error it was on a copied msgbase. So I deleted the entire msgbase and on first pass did not get the error, I only get the error now if I write a message or received msgs in any of the echos or netmail.

If you can reproduce the error in one specific area with the empty
message base, then please send me your full config without passwords
to my email in the tagline.

So empty msgbases do not produce the error during hpt scan etc, when mail arrives in a echo then error appears.


----
Regards,


Rick Smith (Nitro)

... BBSing, the 'Electronic High'
--- GoldED+/LNX 1.1.5-b20180707
* Origin: ----> Abacus Sysop Point --->>>>bbs.abon.us:2323 (1:105/10.1)

Hello Rick,

A 13:52:51 SMAPI ERROR: wrongly sized subfield occured!

If you can reproduce the error in one specific area with the
empty message base, then please send me your full config without
passwords to my email in the tagline.

You sent me one file of your configuration but it contains several files. I asked you to send me your FULL config without passwords. That means that I don't see the error with my configuration, and I need your configuration to reproduce the error. It is impossible with just one file. I need all of them. And the important question is what was before the error occurred in an echo: the echo had no message base, had an empty message base, or the message base contained some messages?

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20180707
* Origin: Moscow, Russia (2:5020/1042)

Hello Rick!

08 Apr 21, Rick Smith wrote to All:

SMAPI ERROR: wrongly sized subfield occured!

Thoughts?

The error report is too short to find a hook how to continue.

Did you extended the loglevel to max? If yes and no clue is found which subfield is effected the next step i would like to wish for is the output of <subfield-name> in the logfile.

You tried my first thought already. Move the .sqd and let the tosser re-create it from scratch. If that fails move the .sqd and try to add a mail to the non-existing area with golded. Then delete the mail with golded and then run the tosser and see if the error is still there. If all that doesn't work there are many possible ways to mismatch a field size.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Greetings Kai!

Saturday April 10 2021 14:06, you wrote to me about an urgent matter!:

* Forwarded from area 'fidosoft.husky'
Hello Rick!

08 Apr 21, Rick Smith wrote to All:

SMAPI ERROR: wrongly sized subfield occured!

Thoughts?

Michael was working with me through email sorry I didnt update here. What I did to fix it was in my huskymak.cfg I changed

PERL=1
USE_HPTZIP=0
DYNLIBS=0

and I was building with the sudo make all install

I used

make all
sudo make install

then just to be safe I deleted all traces of msgbase (this was my point so not horrible to lose the msgs)

And voila it now works as expected. I am curious tho because I have installed this many times and put my backup in place and have never had this problem. Do you suppose it could be manjaro that causes the issues somehow?

anyway thanks for the help everyone...


----
Regards,


Rick Smith (Nitro)

... BBSing very long distance - from Atascadero, California
--- GoldED+/LNX 1.1.5-b20180707
* Origin: ----> Abacus Sysop Point --->>>>bbs.abon.us:2323 (1:105/10.1)

Hello Rick!

10 Apr 21, Rick Smith wrote to Kai Richter:

What I did to fix it was in my huskymak.cfg I changed

PERL=1
USE_HPTZIP=0
DYNLIBS=0

and I was building with the sudo make all install
And voila it now works as expected.

Thanks for the update!

I am curious tho because I have installed this many times and put my backup in place and have never had this problem. Do you suppose it
could be manjaro that causes the issues somehow?

Not directly. Dynlibs puts several functions into external lib files. Those are installed to a lib/ directory which is user-readonly.

I'm not sure what kind of output you'll get on userlevel for

make all
sudo make install

but with dynlibs=1 this will fail because the lib must be installed before the program can be compiled. See the husky base module huskybse files README and INSTALL for detail.


Depending on the *nix distribution there could be specific optimised compiler settings. If those settings change it's always helpful to re-compile those libs too if you want to re-compile software that depends on those libs.

Those dependencies can be very tricky and sometimes it's important to use the correct sequence. E.g. when you use old software that need an old lib version and new tools that need a new version.

The advantage is that the functions inside the lib need to be in memory (RAM and HD) only once - but all linked software can use it.

With dynlibs=0 all required functions of the lib are included directly into the program. If another program uses the same function it would be loaded into memory again and is a RAM wasting dupe.

But even for the raspi this is not relevant for fidonet or husky software. The software is small and there are 3 to ~10 programs running at the same time usually. I think most systems could handle that easily.

If you use dynlibs you should make sure to backup the matching libs too. Without dynlibs your program would work with the single program file.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)

Hello Kai,

Sunday April 11 2021, Kai Richter wrote to Rick Smith:

I'm not sure what kind of output you'll get on userlevel for

make all
sudo make install

but with dynlibs=1 this will fail because the lib must be installed
before the program can be compiled.

No, it will not fail. First one builds and installs huskylib using these two commands. It is important to build any program or library being a regular user, not the root. After that huskylib is installed, so one may build and install smapi using the same two commands. And so on.

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20180707
* Origin: Moscow, Russia (2:5020/1042)

Greetings Michael!

Sunday April 11 2021 18:36, you wrote to Kai Richter about an urgent matter!:

Hello Kai,

Sunday April 11 2021, Kai Richter wrote to Rick Smith:

I'm not sure what kind of output you'll get on userlevel for

make all
sudo make install

but with dynlibs=1 this will fail because the lib must be
installed before the program can be compiled.

Ok, so some further updates. 3 rpi's 3 pc's all running 64bit manjaro. All built with same hpt/binkd setups, same backup ie configs and such with existing msgbase applied to all.

3 pc's all work perfectly and as you would expect, logs and hpt scan free of errors.

3 rpi's all work initially with the backup applied, can read existing messages via golded hpt scan runs without the smapi error.

all 3 pi's have the smapi error once new mail received via poll. the 3 pc's after receiving mail via poll still operate with no smapi error.

So while I thought maybe manjaro was causing me issues, its apparently not manjaro per say but perhaps its manjaro/rpi?


----
Regards,


Rick Smith (Nitro)

... Taglines: the bumper stickers of BBSing.
--- GoldED+/LNX 1.1.5-b20180707
* Origin: ----> Abacus Sysop Point --->>>>bbs.abon.us:2323 (1:105/10.1)

Hello Rick,

Sunday April 11 2021, Rick Smith wrote to Michael Dukelsky:

Ok, so some further updates. 3 rpi's 3 pc's all running 64bit manjaro.
All built with same hpt/binkd setups, same backup ie configs and such
with existing msgbase applied to all.

3 pc's all work perfectly and as you would expect, logs and hpt scan
free of errors.

3 rpi's all work initially with the backup applied, can read existing messages via golded hpt scan runs without the smapi error.

all 3 pi's have the smapi error once new mail received via poll. the
3 pc's after receiving mail via poll still operate with no smapi
error.

So while I thought maybe manjaro was causing me issues, its apparently
not manjaro per say but perhaps its manjaro/rpi?

I think we should not blame the OS, rather it is maybe necessary to update the compiler options. I've never compiled code for Raspberry Pi, so I'll try to guess. :)

Please try to append the flags

-march=armv8-a -mtune=cortex-a72

to OPTCFLAGS in your huskymak.cfg if you have RPi 4 and

-march=armv8-a -mtune=cortex-a53

if you have RPi 3. Also try to set Perl=0 if you don't use Perl and DYNLIBS=1. I am also not sure about LIBDIR=$(PREFIX)/lib in your huskymak.cfg. In my OS all main libraries are in /usr/lib64 and not in /usr/lib. So if it is the same in your OS, you need to use LIBDIR=$(PREFIX)/lib64.

Michael

... node (at) f1042 (dot) ru
--- GoldED+/LNX 1.1.5-b20180707
* Origin: Moscow, Russia (2:5020/1042)

Hello Michael!

11 Apr 21, Michael Dukelsky wrote to Kai Richter:

make all
sudo make install

but with dynlibs=1 this will fail because the lib must be
installed before the program can be compiled.

No, it will not fail. First one builds and installs huskylib using
these two commands.

Sorry, i forgot that "make all" is make all targets within one husky module.

I confused this with make all modules.

Regards

Kai

--- GoldED+/LNX 1.1.4.7
* Origin: Monobox (2:240/77)