• Re: XLOG broken

    From Christoph Berg@21:1/5 to All on Thu Dec 1 22:10:01 2022
    Re: Ed Lawson
    I thought I would post here before filing a formal bug report in case
    this is a known issue and/or workaround available. Today I did a dist-upgrade on a Sid system in shack and neither XDX nor XLog would
    start afterwards. Seg faults on both. Installed XLog on my non-shack Sid system and got same result.

    This isn't a known problem, can you report a bug?

    I know this problem is happening on the Sid tree, but still. Installs
    without any complaint and then simply will not run due to a seg fault? Shouldn't this have been caught in experimental?

    Experimental is only used for new/experimental/beta releases. This is
    a package that has already been around forever.

    Unfortunately, GUI apps are hard to test or else autopkgtest could
    have detected it.

    Christoph DF7CB

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ed Lawson@21:1/5 to All on Thu Dec 1 21:20:01 2022
    I thought I would post here before filing a formal bug report in case
    this is a known issue and/or workaround available. Today I did a
    dist-upgrade on a Sid system in shack and neither XDX nor XLog would
    start afterwards. Seg faults on both. Installed XLog on my non-shack Sid
    system and got same result.

    I know this problem is happening on the Sid tree, but still. Installs
    without any complaint and then simply will not run due to a seg fault? Shouldn't this have been caught in experimental?

    --
    Ed Lawson
    Ham Callsign: K1VP
    PGP Key ID: B80FC40D
    PGP Key Fingerprint: 1E3D C1AD AB42 AFAE 2E07 A3F6 F498 6F10 B80F C40D

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From tony mancill@21:1/5 to Christoph Berg on Thu Dec 1 22:30:01 2022
    Hi Ed, hi Christoph:

    On Thu, Dec 01, 2022 at 10:08:27PM +0100, Christoph Berg wrote:
    Re: Ed Lawson
    I thought I would post here before filing a formal bug report in case
    this is a known issue and/or workaround available. Today I did a dist-upgrade on a Sid system in shack and neither XDX nor XLog would
    start afterwards. Seg faults on both. Installed XLog on my non-shack Sid system and got same result.

    This isn't a known problem, can you report a bug?

    I'm seeing the segfault too. I'll create a bug and take a look.

    I know this problem is happening on the Sid tree, but still. Installs without any complaint and then simply will not run due to a seg fault? Shouldn't this have been caught in experimental?

    Experimental is only used for new/experimental/beta releases. This is
    a package that has already been around forever.

    Unfortunately, GUI apps are hard to test or else autopkgtest could
    have detected it.

    Cheers,
    tony KG7IEL

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEE5Qr9Va3SequXFjqLIdIFiZdLPpYFAmOJG88ACgkQIdIFiZdL Ppa8mhAA0jm2tf9OMBhvPwPrQAi8AETYXbJxMN17dFpOziaWRqxgeYzIIOPUuxzS ZOrPm1j2o9V7aA8n+YXY4iws1rF6/CH5MQm9w0nF8LRLawofaCOV4MPqr+JXjZnl OwSM60/zDfFEZuIPM4wZP9zA5yQvoJ7pXX8D2zjxN0s97aiAjpHD/maRY17PV+LL 9n889v9svMk+98IadKXQDx7bcBAEEfmzzA9gGXB6FhYb5aQKP3S371uOSsP0m5NT IoAHpvHRjsxd8Kx5FgHd3XPaelnk062RTH1IQzU/+oYMM4jhLAF+LzWANuEcxsoC /HzY1TEAHxYq7VOGkZaT1TxjaEd0T7m84IlaN7m72TKzB4I1GvGyuh7i5Dp3cKNx vj19pH0oiKelyiLxUC3SyczZJXagbbjBCPG443RQYXgG6x0D9TWC8y1/QogGjgsk MOx4ma7ESqFofnrDVdZOEMTQujbG/nsRSKv12X0kKyMlkIbRXX60CVS3O0I3ZgHs xk7Axy11sqPlcBkwew3JxmwIN9DPCETPhD3AfuGgGakDm8NKGNRgIKe2u2YgZNT4 1TXLyMZ4uw9ThOo9xwEU33Be+mRKqeMocxY9CWPys8Q3334aAidjRsybJWww7Mzz lQpLvjmB86W/uGAiyUHhjNksAgQu3OtzGBY+rUGuHsnbZ62qqCo=
    =O3bt
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ed Lawson@21:1/5 to All on Fri Dec 2 03:10:01 2022
    Tony:

    Thank you for filing the bugreport. It seems xdx is also failing to
    open due to a seg fault for other reasons, and I have filed a bugreport
    for it.

    For xlog the cause seems to a failure to load a module for acessibility.

    --
    Ed Lawson
    Ham Callsign: K1VP
    PGP Key ID: B80FC40D
    PGP Key Fingerprint: 1E3D C1AD AB42 AFAE 2E07 A3F6 F498 6F10 B80F C40D

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From tony mancill@21:1/5 to Ed Lawson on Fri Dec 2 05:40:01 2022
    On Thu, Dec 01, 2022 at 08:32:11PM -0500, Ed Lawson wrote:
    Tony:

    Thank you for filing the bugreport. It seems xdx is also failing to
    open due to a seg fault for other reasons, and I have filed a bugreport
    for it.

    I will try to take a look at this weekend.

    For xlog the cause seems to a failure to load a module for acessibility.

    In that case, there might be multiple bugs. I was also seeing a
    segfault on startup, but it was consistently due to an insufficiently
    sized buffer while reading cty.dat. Increasing the buffer size
    (doubling it) resolved the issue for me, although it's not very
    satisfying. First, I haven't identified exactly why this started
    happening - perhaps the recent updates to glib2.0 [1]? - so I have some follow-up to do there. And second, the fixed buffer size could cause
    problems in the future, although that's more of an upstream issue.

    If xlog 2.0.24-3 still segfaults for you, please do open another bug.
    It's working for me locally now.

    Cheers,
    tony

    [1] https://tracker.debian.org/pkg/glib2.0

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEE5Qr9Va3SequXFjqLIdIFiZdLPpYFAmOJgK4ACgkQIdIFiZdL Ppal0hAAk/SO6RwbjAEls+vmNNXmjtbs1xNWld4k+dU5tMeVeLzsBRXiZFa4R35E qxindak3q29wIXrfixbXdgW4SA8dFYiKdx+Mtp8UpdloL3EiktE1PooACd6jAaox iX8cJFxesNbF/Ox5eMCTbRWUnqQZAh/CzaDGKVAgffd4ij9AtMmUJk/561jG8oF3 KlSnySYwQG0dZM2gIjI954Cw62nzonISNXUgmv+qFU6z8npREBZPq5sgGuD0bpLx yFY5HDPXx9epnWK7QyS07frYh6T4qsyhACXed3E6hBXwCpdSbzOJyMYmo4kb1kYB bEBv0XnstfeuTRgTDX5qDV94u6dc3atXOnfiDsPB/eOIxR/lc4C4iY6mLPs3hQM+ HNziHCh8aCnIsqQBLAgMOhMKjCgP1Nf6C2Ql6xZISvQp/afo537CpnnWJ6ImTMag qcoXSByAUBlg5UWLDjVlcV9SrjwGiUznikpibKq0YrS7ygdiE9XSfuaPGJlmVHGs XwYyxP1iaoVHz3g9gt7aNZD49K3H1fnzC0cTAUJEqxBTToNKaCLCY9jSwaC24gWv zBrJRR+5Jvgz/zjjNyCVcaOedtV80XbHnly6RxzdV9pQ/ErPzHK5iEfLbtnlXQ4F 4HKRMWU4qrI7lBHHpZaOL3eC4NvbRluPnI4Akg+ryALMcdl4BZg=
    =HbVm
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christoph Berg@21:1/5 to All on Fri Dec 2 11:50:02 2022
    Re: tony mancill
    In that case, there might be multiple bugs. I was also seeing a
    segfault on startup, but it was consistently due to an insufficiently
    sized buffer while reading cty.dat. Increasing the buffer size
    (doubling it) resolved the issue for me, although it's not very
    satisfying. First, I haven't identified exactly why this started
    happening - perhaps the recent updates to glib2.0 [1]?

    Perhaps the new cty.dat from hamradio-files has more data in it - some countries have very long lists of callsigns.

    Christoph

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Ed Lawson@21:1/5 to tony mancill on Fri Dec 2 15:20:01 2022
    On Thu, 1 Dec 2022 20:35:59 -0800
    tony mancill <[email protected]> wrote:



    If xlog 2.0.24-3 still segfaults for you, please do open another bug.
    It's working for me locally now.


    Tony:

    xlog 2.0.24-3 working FB here. Thank you.
    --
    Ed Lawson
    Ham Callsign: K1VP
    PGP Key ID: B80FC40D
    PGP Key Fingerprint: 1E3D C1AD AB42 AFAE 2E07 A3F6 F498 6F10 B80F C40D

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From tony mancill@21:1/5 to Christoph Berg on Fri Dec 2 22:00:01 2022
    On Fri, Dec 02, 2022 at 11:40:40AM +0100, Christoph Berg wrote:
    Re: tony mancill
    In that case, there might be multiple bugs. I was also seeing a
    segfault on startup, but it was consistently due to an insufficiently
    sized buffer while reading cty.dat. Increasing the buffer size
    (doubling it) resolved the issue for me, although it's not very
    satisfying. First, I haven't identified exactly why this started
    happening - perhaps the recent updates to glib2.0 [1]?

    Perhaps the new cty.dat from hamradio-files has more data in it - some countries have very long lists of callsigns.

    Thank you for helping me make the connection. (That makes a lot more
    sense than the complicated scenarios I was imagining... :) I was thrown
    off the trail by the data/dxx/cty.dat in the xlog source package and
    had assumed that it was what was being read (and knew that it hadn't
    changed).

    But I see now:

    $ find /usr/share -name cty.dat -exec ls -al {} \;
    -rw-r--r-- 1 root root 321922 Nov 25 10:54 /usr/share/hamradio-files/cty.dat lrwxrwxrwx 1 root root 25 Aug 16 04:42 /usr/share/xdx/cty.dat -> ../hamradio-files/cty.dat
    -rw-r--r-- 1 root root 79386 Aug 27 09:56 /usr/share/tucnak/cty.dat
    lrwxrwxrwx 1 root root 25 Jul 31 21:08 /usr/share/tlf/cty.dat -> ../hamradio-files/cty.dat
    lrwxrwxrwx 1 root root 28 Oct 17 2021 /usr/share/xlog/dxcc/cty.dat -> ../../hamradio-files/cty.dat
    lrwxrwxrwx 1 root root 25 Nov 23 02:11 /usr/share/wsjtx/cty.dat -> ../hamradio-files/cty.dat

    And there have been plenty of recent updates to cty.dat in hamradio-files [1]

    Thanks again to Ed for the bug report!

    [1] https://salsa.debian.org/debian-hamradio-team/hamradio-files/-/commits/master/bigcty/cty.dat

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEE5Qr9Va3SequXFjqLIdIFiZdLPpYFAmOKZbIACgkQIdIFiZdL PpZVlA//eBtu8buMipAo7+V8KyXKnBijLrieSOmizlYEJ5Lk62K4CFxy/M0lHhcF 9p69lFJD9RLBExL8YmugIbjE/lpYc2bQsuigWie4e9LRDuT9QDBEGPZRsh5MnIyR hLacAjZ9nSm6XjHATcbRmgQWWLZqhllr/h/3++zLqVjEzDOGWcmEYN1bMPAAPvbL UkR8pYjG5bViC1cDKGWue7eNcdO9SeDr1pSbqGQsouhP9Gc6Jxfi8CYuKmkGHDs5 n2n0PM3uK/Q0qDUXd6V5aFRu3eVVlDMfgdT/JPaXUZQrqTvbKEaSvFqp3xwuSf2F G4isQ0N7zc25TDW/TrftqR0ZvX8m+y2Ly9C5+3gjiA2k8wT00x8mDLt278mbxYVG RWCC2zKxRMBL6Nx0KTJLsOV0QyR8LOY3vWacbeSZA81ZHzmnwhrArSvOvU83i9CU dZcs0LB1sC/T/jJxDD7Tb19kkDO4p6SYFLDxJrPBAbEiWJin7f86nFKNk6nQeTqh BZV7d8suikcZQnsxqtBN9P7C9gNYx5/DHdhDj3a6rS6A/+RU08EZ4BxFX2tHG8qS QXop6KCi6Uh56nSBRAAaVVt71jswjkEv6iUOMapB2DXcNfZMeKqrRb6q64TNHC6c pTjgrYAxNK1D8Sod8MkOJ+AWM6fRNeFSwJ2XDCHLZ+BE699zb/0=
    =AbxZ
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christoph Berg@21:1/5 to All on Sun Dec 4 14:30:01 2022
    Re: tony mancill
    Thank you for helping me make the connection. (That makes a lot more
    sense than the complicated scenarios I was imagining... :) I was thrown
    off the trail by the data/dxx/cty.dat in the xlog source package and
    had assumed that it was what was being read (and knew that it hadn't changed).

    There are two versions of the country files, a small one with just a
    few callsign exceptions, and the "bigcty" one with the long list that
    we are using now. xdx had assumed the file is smaller (while still
    having the bigcty version in mind, I think):

    /* Buffer size for reading in a single cty.dat record. Currently the
    * entire cty.dat file is just under 80k bytes, so any single record should
    * be much less than this value, but with additions of callsign exceptions
    * record sizes will undoubtedly continue to grow.
    */
    #define MAX_RECORD_SIZE 65536

    -rw-r--r-- 1 myon myon 288472 25. Nov 03:38 bigcty/cty.csv
    -rw-r--r-- 1 myon myon 325766 25. Nov 03:38 bigcty/cty.dat

    I'm adding a check to hamradio-files to see if our new assumption of
    not exceeding 128kB holds:

    https://salsa.debian.org/debian-hamradio-team/hamradio-files/-/commit/8af7facd536485e56d86bac10d214108f2e0c075

    Christoph

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From tony mancill@21:1/5 to Christoph Berg on Sun Dec 4 18:10:01 2022
    On Sun, Dec 04, 2022 at 02:28:03PM +0100, Christoph Berg wrote:
    Re: tony mancill
    Thank you for helping me make the connection. (That makes a lot more
    sense than the complicated scenarios I was imagining... :) I was thrown off the trail by the data/dxx/cty.dat in the xlog source package and
    had assumed that it was what was being read (and knew that it hadn't changed).

    There are two versions of the country files, a small one with just a
    few callsign exceptions, and the "bigcty" one with the long list that
    we are using now. xdx had assumed the file is smaller (while still
    having the bigcty version in mind, I think):

    /* Buffer size for reading in a single cty.dat record. Currently the
    * entire cty.dat file is just under 80k bytes, so any single record should
    * be much less than this value, but with additions of callsign exceptions
    * record sizes will undoubtedly continue to grow.
    */
    #define MAX_RECORD_SIZE 65536

    -rw-r--r-- 1 myon myon 288472 25. Nov 03:38 bigcty/cty.csv
    -rw-r--r-- 1 myon myon 325766 25. Nov 03:38 bigcty/cty.dat

    I'm adding a check to hamradio-files to see if our new assumption of
    not exceeding 128kB holds:

    https://salsa.debian.org/debian-hamradio-team/hamradio-files/-/commit/8af7facd536485e56d86bac10d214108f2e0c075

    Good idea! Let's see how long it takes us to exceed 128kB.

    $ wc --max-line-length hamradio-files-20221125/bigcty/cty.csv
    67779 hamradio-files-20221125/bigcty/cty.csv

    For those curious about the longest lines...

    $ cat hamradio-files-20221125/bigcty/cty.csv | awk '{ print length, $0 }' | cut -d',' -f1-2 | sort -rn -s | head -10
    67780 K,United States
    43579 UA9,Asiatic Russia
    27662 BY,China
    24590 LU,Argentina
    18650 UA,European Russia
    7790 GM,Scotland
    6818 GW,Wales
    5564 KL,Alaska
    5208 KH6,Hawaii
    3776 VE,Canada


    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEE5Qr9Va3SequXFjqLIdIFiZdLPpYFAmOM0ncACgkQIdIFiZdL PpZTCBAAicGH8qKmBuAevyQsDqZmHghs2+oPxVnEmDJwu0yfvhsXEK6RdHS6l6Ln 8zL2lN/CH2xohpBS8KrQHj+thIEvSP/8XNfKip7zitMD5+UvJ8CPMqKxEBYfSb6p Ue5y/4aRFugTg566puvJDIU+qeFWIqzIH6KUR21/izTTYqJeO7ioOkz+qF8YZAQf ubjx8d3pMDPbSgjFZ8BVU/8qsZJ9HdZb7KbqJYDK+AeFiwerPVRgBl1BJDkMwXdf J9Z776S3LkLCsu2ObcdzFz7Td2gO1xBhUakqp4MzAq+hw52lRRaWwUGpXaCJfK3Z 9YzE+NfLmDZeLEcYYENjFi9Bc6aOzKz7jp5v1VupWq0f07bW2wedQvt6VZ0allC1 LCigdB7sAMZn1saAwMq/nf5nCuyXLZgimj9Z3gchmoM0/3Upl94dBTXiMw6KubjP wDAJ664JdpJvcmc4mDvGSMww2jVb9wiQEsIOdXOD4xkLtBL6seELFgJOo/5T6Nwd nQVh56XGkTADZKL4FvUfwwirC+q7Pvmo0FV3+d+fXPTdTEGJBvua41biOVszuFIm CKeT/IB5tJiRfg6DDd8HNeAk/UwE0q+4QOQwsFuXAFnioQWxIRrAenKY8srxUAkw J5O8rrLMo1rToEvThVkknCPBUDWmw13c5nNFfrIyTRHdH5OwTaU=
    =n7Rq
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christoph Berg@21:1/5 to All on Sun Dec 4 20:20:01 2022
    Re: tony mancill
    For those curious about the longest lines...

    $ cat hamradio-files-20221125/bigcty/cty.csv | awk '{ print length, $0 }' | cut -d',' -f1-2 | sort -rn -s | head -10
    67780 K,United States
    43579 UA9,Asiatic Russia
    27662 BY,China
    24590 LU,Argentina
    18650 UA,European Russia
    7790 GM,Scotland
    6818 GW,Wales
    5564 KL,Alaska
    5208 KH6,Hawaii
    3776 VE,Canada

    And for the "why", K/UA9/BY/LU/UA/VE are the countries spanning
    multiple CQ/ITU zones, and GM/GW/K* have lots of stations situated in
    the "wrong" part of the country. Additionally, LU has a shitload of
    exceptions for callsigns of the form LU1xx/D and other suffixes that
    seem to be some internal thing, not indicating a different country.

    Christoph

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)