• Odd Golded Question?

    From mark lewis@1:3634/12.73 to all on Sat Dec 29 09:03:38 2018
    On 2017 Jan 27 11:08:12, Rick Christian wrote to Jeff Smith:

    yes, i know this is old and i'm top posting... sorry...

    i've spent the last day hunting down this patch because my golded+ was crashing
    with "stack smashing detected"... yeah, i was running my terminal window at max
    screen size which is now 271x70... if i drop to 256x70, i can read but, as noted below, the internal editor craps the sheets... dropping back to 255x70 allows both to work... i am NOT using the patch... it didn't seem to do any good either way and i was still limited in max COLSxROWS... i even added a second zero to make the buf larger and that didn't help, either... the problem is the same with my old golded using the code from cvs and this new one using the latest code from github... it is really starting to look like there's a byte size counter somewhere that is the real problem... i hate that my C/C++ skills are non-existant these days :(


    I am too impatient to enlarge, start Golded, rinse, repeat to find the limit where things break...

    BUT

    On my 26-50" screens full screen terminals will break golded starting, period,

    THere is a patch: https://fidonet.ozzmosis.com/echomail.php/golded/37e10bdf0c33c539.html

    The patch below fixes a bug where a very wide terminal causes GoldED to crash:

    ---------- Fri 21 Nov 14, GoldED+/BSD 1.1.5-b20130910 (Nov 16 2014 22:19:02) ! 23:02:48 Memory error at [gemlst.cpp,365]. ! 23:02:48 gsprintf(buffer,256,%-5.5s%s%-*.*s %-*.*s%s%-*.*s %s,...): buffer overflow (need 381 bytes).


    Index: gemlst.cpp ===================================================================
    RCS file: /cvsroot/golded-plus/golded+/golded3/gemlst.cpp,v
    retrieving revision 1.67
    diff -u -r1.67 gemlst.cpp
    -+- gemlst.cpp 19 Oct 2011 23:31:31 -0000 1.67
    +++ gemlst.cpp 22 Nov 2014 16:50:45 -0000
    {at}{at} -333,7 +333,7 {at}{at}
    mattr_ = hattr;
    }

    - char buf[256];
    + char buf[2560];

    if(AA->Msglistwidesubj()) {
    resiz += tosiz + 1;


    Which does let wide terminals work... I've done it to all the ones of recent... and works.

    BUT

    ONLY FOR READING

    If you reply the internal editor barfs and seg faults... for probably similar reasons... but I just don't have time to dig through to find its buffer to enlarge it.

    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... When I spilled my cottage cheese, I lost my whey.
    ---
    * Origin: (1:3634/12.73)
  • From mark lewis@1:3634/12.73 to mark lewis on Sat Dec 29 09:21:12 2018
    On 2018 Dec 29 09:03:38, I wrote to all:

    On 2017 Jan 27 11:08:12, Rick Christian wrote to Jeff Smith:

    yes, i know this is old and i'm top posting... sorry...

    i've spent the last day hunting down this patch because my golded+ was crashing with "stack smashing detected"...


    i forgot to post that...


    ----- snip -----
    Pick New Area: ÚÄGÄDescriptionÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄEchoIDÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄNewÄÄÄMsgsÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ¿
    ³ÄÄÄNetmail areasÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ ÄÄÄij
    ³ NetMailArea NetMailArea
    0 1071 *** stack smashing detected ***: /home/wkitty42/fido/bin/gedlnx terminated ³
    ³ Aborted (core dumped) ³
    ----- snip -----


    it always gets the netmail area, as above, and then dies... 255x70 works, 256x70 does not...


    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... URA True Northerner if sittin on yer deck shootin ravens is fun.
    ---
    * Origin: (1:3634/12.73)
  • From andrew clarke@3:633/267 to mark lewis on Mon Dec 31 06:29:42 2018
    29 Dec 18 09:03, you wrote to all:

    i've spent the last day hunting down this patch because my golded+ was crashing with "stack smashing detected"... yeah, i was running my
    terminal window at max screen size which is now 271x70... if i drop to 256x70, i can read but, as noted below, the internal editor craps the sheets... dropping back to 255x70 allows both to work... i am NOT using the patch... it didn't seem to do any good either way and i was still limited in max COLSxROWS... i even added a second zero to make the buf larger and that didn't help, either... the problem is the same with my
    old golded using the code from cvs and this new one using the latest
    code from github... it is really starting to look like there's a byte
    size counter somewhere that is the real problem... i hate that my C/C++ skills are non-existant these days :(

    I wrote the first patch for that back in 2014 but it was never accepted upstream.

    Consequently the current version of GoldED still has that bug.

    I never tested >255 char wide screens with the editor, partly because my fix wasn't accepted upstream, so I forgot about submitting any more patches, but mainly because I never use terminals that large and I didn't expect anyone else
    to.

    See the next series of messages for my fixes to the bugs I know about. Be warned I haven't tested them extensively with very large terminals so you're likely to encounter more crashes.

    --- GoldED+/BSD 1.1.5-b20180707
    * Origin: Blizzard of Ozz, Melbourne, Victoria, Australia (3:633/267)
  • From mark lewis@1:3634/12.73 to andrew clarke on Sun Dec 30 16:21:38 2018
    On 2018 Dec 31 06:29:42, you wrote to me:

    I wrote the first patch for that back in 2014 but it was never
    accepted upstream.

    yeah, i know/remember... they were the first to show up in my local searches...
    i was actually looking for a simple golded compilation script that i thought someone had posted but i wasn't able to find what i thought i remembered... not
    even after spending a few hours digging back through ozzmosis...

    Consequently the current version of GoldED still has that bug.

    yep... funny thing is, once i remembered how to compile golded and got the latest code, i was still seeing the "stack smashing" crash... it wasn't until is suddenly dawned on me that i had rebooted the system and this new wide screen monitor was now better recognized that i got my first clue about the terminal width being the problem... so i went looking for the "[2560]" patch to
    be sure where to put it... it didn't help and i still had the same problem... so i took that patch back out and built again... then i tried dropping my terminal width and loading golded... that's how i figured out the 255 width limit... heighth didn't seem to matter... the real thing that initially threw me off was the error message changed to the "stack smashing" one... that almost
    lead me down the wrong rabbit hole at first...

    I never tested >255 char wide screens with the editor, partly because
    my fix wasn't accepted upstream, so I forgot about submitting any more patches, but mainly because I never use terminals that large and I
    didn't expect anyone else to.

    hehehe... yeah, i spend almost all of my time in one of over a dozen terminal windows i have open at any one time... the majority of them are opened max (fullscreen?)... on this new monitor, that's 271x70 with my current configuration...

    See the next series of messages for my fixes to the bugs I know about.
    Be warned I haven't tested them extensively with very large terminals
    so you're likely to encounter more crashes.

    i will look...

    i am happy to say that something certainly has been corrected since my last code update and build... i wasn't able to do any nodelist lookups but now that seems to work... i think i have a little more work to do to get the nodelist stuff working properly (ie: name lookups) but i'll get to that at some point...
    if anyone has any pointers, they'd be appreciated... right now my nodelist is being compiled into three goldnode.gx[aln] files...

    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... Give up, you'll only live till you die.
    ---
    * Origin: (1:3634/12.73)
  • From mark lewis@1:3634/12.73 to andrew clarke on Sun Dec 30 16:54:54 2018
    On 2018 Dec 30 16:21:38, I wrote to you:

    See the next series of messages for my fixes to the bugs I know
    about. Be warned I haven't tested them extensively with very large
    terminals so you're likely to encounter more crashes.

    i will look...

    NOTE: the 3rd time is the charm? my previous attempts crashed when i hit CTRL-Z
    to bring up the menu to save and exit... luckily i was able to find the below text of my original post in golded.tmp and rescue it before trying again... i'm
    doing this in the known working 255x70 terminal... i'll try 256 in a minute and
    if that works, then larger until it crashes but i think we know what'll happen... i'll also try to remember to grab that crash message this time... might still have the last ones in the terminal scroll back buffer ;)


    i've applied those fixes and am now reading and writing this reply in a maximized terminal window of 217x70... so far, this initial test seems to be ok... let's see what happens when i save this reply and move on to another area... if you get this message, then i guess everything is working as needed ;)

    i guess i should probably update my version number somehow to show that this is
    a custom build of the code i have with your patches applied...


    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... Happy Holidays and a Wonderful 2011 to you and yours!
    ---
    * Origin: (1:3634/12.73)
  • From mark lewis@1:3634/12.73 to andrew clarke on Sun Dec 30 17:07:28 2018
    On 2018 Dec 30 16:54:54, I wrote to you:

    trying again... i'm doing this in the known working 255x70 terminal... i'll try 256 in a minute and if that works, then larger until it
    crashes but i think we know what'll happen... i'll also try to
    remember to grab that crash message this time... might still have the
    last ones in the terminal scroll back buffer ;)

    well, dammit... i've just written a test message, a G reply to that message and
    a Q reply to /that/ message... all at 271x70 and everything works... they're over in FIDOTEST, FWIW...

    i wasn't able to catch those last crash codes... i'll do that the next time it happens...

    how would i run golded in gdb so it can find the proper source code lines? that
    might be complicated by me running golded via the cp437 tool that converts cp437 characters to the proper ones for display on the screen... hummm...

    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... All of my voices listen to different radio stations.
    ---
    * Origin: (1:3634/12.73)
  • From mark lewis@1:3634/12.73 to andrew clarke on Sun Dec 30 17:34:02 2018
    On 2018 Dec 30 17:07:28, I wrote to you:

    i wasn't able to catch those last crash codes... i'll do that the next
    time
    it happens...

    got it! looks to maybe be related to having long lines covering the full screen
    width... if i keep the lines short, no problem... this one is maybe 3/4s of my screen width...
    whatever it is, it seems to happen when i CTRL-Z to save the message... maybe it has something to do with processing golded.tmp and getting the message body from there?

    sooo, on the crash, some information scrolls off the screen and i can't get to it... but i did get a screen shot... it ain't pretty...

    http://sestar.synchro.net/temp/20181230-golded-widescreen-crash-event.jpg

    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... Today's mood: cranky with a touch of psycho
    ---
    * Origin: (1:3634/12.73)
  • From Paul Quinn@3:640/1384 to mark lewis on Mon Dec 31 10:16:42 2018
    Hi! mark,

    On 30 Dec 18 16:21, you wrote to andrew clarke:

    i am happy to say that something certainly has been corrected since my last code update and build... i wasn't able to do any nodelist lookups
    but now that seems to work... i think i have a little more work to do
    to get the nodelist stuff working properly (ie: name lookups) but i'll
    get to that at some point... if anyone has any pointers, they'd be appreciated... right now my nodelist is being compiled into three goldnode.gx[aln] files...

    Those three files are created here as well. I have these entries in my GoldEd.cfg (they are the only nodelist related statements, BTW):

    --- 8< ---
    ; The path to your nodelists.
    ; NOTE: This keyword MUST be in GOLDED.CFG!!!
    NODEPATH /opt/ftn/fido/nodelist/

    ; Filenames of your nodelists.
    ; If extention is ".*" or ".999", the newest is used automatically.
    ; NOTE: This keyword MUST be in GOLDED.CFG!!!
    ; xxx use daily lists from 22 August 2018...
    ; xxxx NODELIST nodelist.999
    NODELIST z3daily.999

    ; Uncomment this to use Version 7 nodelist indexes.
    ;xxxxNODEPATHV7 /opt/ftn/fido/nodelist/

    ; Uncomment this to use a raw FIDOUSER.LST file.
    ;xxxxFIDOUSERLIST fidouser.lst

    ; Where to allow nodelist lookup when entering messages.
    LOOKUPNET YES
    LOOKUPECHO NO
    LOOKUPLOCAL NO
    --- >8 ---

    When the daily nodelist is unpacked, the following code updates the index files:

    --- 8< ---
    cd /opt/ftn/fido/exteditor
    ./gnlnx -force
    --- >8 ---

    But you probably already knew that bit. Oops. :)

    Cheers,
    Paul.

    ... Dark Shadows - https://www.youtube.com/watch?v=KwleloRuaEw
    --- GoldED+/LNX 1.1.5-b20130515
    * Origin: Quinn's Rock - Live from Paul's Xubuntu desktop! (3:640/1384)
  • From mark lewis@1:3634/12.73 to Paul Quinn on Sun Dec 30 20:28:48 2018
    On 2018 Dec 31 10:16:42, you wrote to me:

    to get the nodelist stuff working properly (ie: name lookups) but i'll
    get to that at some point... if anyone has any pointers, they'd be
    appreciated... right now my nodelist is being compiled into three
    goldnode.gx[aln] files...

    Those three files are created here as well. I have these entries in my GoldEd.cfg (they are the only nodelist related statements, BTW):

    i have the same and the same ones commented out, too...

    When the daily nodelist is unpacked, the following code updates the
    index files:

    i'll try to remember that... mine were generated automatically earlier... i guess when GE started or maybe i keyed something that did it from inside GE? i dunno... i'll remember that command, though...

    But you probably already knew that bit. Oops. :)

    i hadn't quite got that far yet :lol: -=B-b

    )\/(ark

    Always Mount a Scratch Monkey
    Do you manage your own servers? If you are not running an IDS/IPS yer doin' it wrong...
    ... Let your life dance lightly on the edges of time like the dew on a leaf. ---
    * Origin: (1:3634/12.73)