On Thursday 04 November 2010 12:15:25 Sergei Golovan wrote: > > On Thu, Nov 4, 2010 at 9:43 AM, Sergei Golovan <sgolo...@nes.ru> wrote: > > On Thu, Nov 4, 2010 at 1:21 AM, Emil Langrock <emil.langr...@gmx.de> wrote: > >>> odbc:start(). > >> > >> odbc:connect("Driver=SQLITE3;Database=/var/lib/ejabberd_archive/mod_arch > >> ive.sqlite3;", [{scrollable_cursors, off}]). > >> > >> =ERROR REPORT==== 3-Nov-2010::22:46:50 === > >> ODBC: received unexpected info: {tcp_closed,#Port<0.670>} > >> > >> > >> =ERROR REPORT==== 3-Nov-2010::22:46:50 === > >> ** Generic server <0.57.0> terminating > >> ** Last message in was {#Port<0.668>,{exit_status,139}} > >> ** When Server state == {state,#Port<0.668>, > >> {<0.35.0>,#Ref<0.0.0.109>}, > >> > >> <0.35.0>,undefined,on,undefined,undefined,off, connecting,undefined,0, > >> [#Port<0.666>,#Port<0.667>], > >> #Port<0.669>,#Port<0.670>} > > > > odbcserver (the Erlang port which talks to unixodbc) segfaults when > > connecting to > > SQLite3 database. Seems like a bug in odbcserver. I'll try to > > investigate it. Though I think the fix will appear in unstable and > > testing first, not in experimental (where your 1:14.b-dfsg-1 comes > > from). > > In fact, the segfault happens in /usr/lib/odbc/libsqlite3odbc.so which > is part of > libsqliteodbc package. > > After some digging in the code I've found that setting connect attribute > SQL_AUTOCOMMIT leads to the segfault. (Perl and python do this other way, > so they work.) > > Appears that the author of sqliteodbc has fixed this bug (see changelog for > version 0.87 at http://www.ch-werner.de/sqliteodbc/). I've extracted a > small patch which make crach gone and attached it to this message. > > So, I'm reassigning this bug to libsqliteodbc. > > Cheers!
Thanks for your debugging. It works a lot better with 0.87, but still seems to kill itself when I try to write (probably write). I only tested ejabberd with mod_archive_webview. In previous sqliteodbc versions it was not possible to see the browse, search or use any other submenu. Now i can write (xmpp) to an account with globally enabled xep 0136 logging through ejabberd mod_archive_odbc and get: =CRASH REPORT==== 4-Nov-2010::17:25:04 === crasher: initial call: odbc:init/1 pid: <0.458.0> registered_name: [] exception exit: {port_exit,killed} in function gen_server:terminate/6 ancestors: [odbc_sup,<0.285.0>] messages: [{'EXIT',#Port<0.3714>,normal}] links: [<0.286.0>] dictionary: [] trap_exit: true status: running heap_size: 1597 stack_size: 24 reductions: 2104 neighbours: =SUPERVISOR REPORT==== 4-Nov-2010::17:25:04 === Supervisor: {local,odbc_sup} Context: child_terminated Reason: {port_exit,killed} Offender: [{pid,<0.458.0>}, {name,[]}, {mfargs,{odbc,start_link_sup,[[{client,<0.277.0>}]]}}, {restart_type,temporary}, {shutdown,7000}, {child_type,worker}] -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org