:: #mythtv

Daily chat history

Current users (65):

aloril, amessina, Anssi, brfransen, caelor, cesman, Chutt, clever, coling, dblain, eee-blt_, ElmerFudd, enyc, esperegu, fetzerch, Gibby, GreyFoxx, J-e-f-f-A, jams, jarle, jheizer, joki, jpharvey, jst, jwhite, jya, knightr__, kormoc, kurre2, moparisthebest, MythBuild, MythLogBot, nephyrin, og01_, peper03, poptix, purserj, rhpot1991, rich0, rmeden, robink, rsiebert_, Seeker`, seld, Sharky112065, sheedy-away, skd5aner, sl1ce, sphery, sraue, stuartm, suffice, supaplex, superm1, taylorr, tgm4883, tonsofpcs, tris, unforgiven512, wagnerrp, Warped, XDS2010, xris, zentec_, _charly_
Monday, October 6th, 2014, 00:02 UTC
[00:02:26] stuartm: new size is wrong in the realloc, should be "pkt->size + grow_by + FF_INPUT_BUFFER_PADDING_SIZE;" but the supplied input is 384 + 2672 + 32 which does not equal 3088
[00:03:16] stuartm: no it does ... heh, I really should get some sleep
[00:17:41] stuartm: jya: it doesn't crash if I replace the code in set_pkt with av_copy_packet
[00:18:15] jya: stuartm: the idea was to avoid copying…
[00:18:35] jya: actually, that was one of the proposal elsewhere.. but that was leaking
[00:18:58] jya: i may get a look at it later today… what argument are you providing to mythtranscode so I can reproduce it here ?
[00:19:07] stuartm: right, just noting that although the crash occurs later in ensure_size, the root issue is set_pkt managing to produce an invalid av_packet
[00:19:22] stuartm: mythtranscode --loglevel debug -m -i /mnt/myth1/recordings/23208_20140722175900.mpg -o transcode_test.mpg
[00:20:20] jya: ok… will trye
[00:20:46] stuartm: the result with av_copy_packet is perfect
[00:45:00] coling (coling! has quit (Ping timeout: 260 seconds)
[00:56:32] Warped (Warped!~Warped@ has quit (Ping timeout: 245 seconds)
[01:01:33] jya: stuartm: I’ll look into it now
[01:02:49] Warped (Warped!~Warped@ has joined #mythtv
[02:17:15] peper03_ (peper03_!~peper03@mythtv/developer/peper03) has joined #mythtv
[02:21:08] peper03 (peper03!~peper03@mythtv/developer/peper03) has quit (Ping timeout: 260 seconds)
[02:21:08] peper03_ is now known as peper03
[02:29:19] arescorpio (arescorpio!~arescorpi@ has joined #mythtv
[02:36:12] gigem (gigem!~david@mythtv/developer/gigem) has quit (Quit: WeeChat 1.0)
[02:37:09] gigem (gigem! has joined #mythtv
[02:37:09] gigem (gigem! has quit (Changing host)
[02:37:09] gigem (gigem!~david@mythtv/developer/gigem) has joined #mythtv
[03:48:24] Warped (Warped!~Warped@ has quit (Ping timeout: 250 seconds)
[03:50:58] fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has quit (Ping timeout: 260 seconds)
[03:52:09] fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has joined #mythtv
[04:07:03] Warped (Warped!~Warped@ has joined #mythtv
[04:54:54] arescorpio (arescorpio!~arescorpi@ has quit (Excess Flood)
[05:45:14] dekarl: paul-h, thanks for going through the list. But the infoneeded -> new switch on wasn't so good. The user did'nt do what I asked him to do and did not provide any helpful information... He even stats how nice he is logging but not providing anything... meh
[06:15:01] dekarl: oh, now I see 12133 has been resolved, nice :)
[06:15:24] SteveGoodey (SteveGoodey! has joined #mythtv
[06:17:28] FabriceMG (FabriceMG! has joined #mythtv
[06:22:00] dekarl: re #12047 what do our gurus think about using the ffmpeg helper for guessing the video frame rate instead of guessing ourselves? It doesn't help in that case, but is a step to reduce code duplication. . . . lace-o.patch
[06:22:00] ** MythLogBot **
[06:23:38] dekarl: in the example file, only the codec_fps gets the expected value of 25 fps, . . . mkv.log#L167
[06:26:22] jya: dekarl: you’ll want to look at 2.4 , I;m planning to resync this week. (ffmpeg dropped support to 2.3 already)
[06:32:51] jya: stuartm: ok.. I can reproduce the crash finally…. working on it
[06:53:31] Tobbe5178 (Tobbe5178! has joined #mythtv
[06:53:34] jya: found it
[06:57:49] coling (coling! has joined #mythtv
[08:01:11] stuarta: morning all
[08:03:14] jya: morning!
[08:06:37] Merlin83b (Merlin83b!~Daniel@2a00:1ee0:3:1337:7506:e5c7:b8be:c1cf) has joined #mythtv
[08:17:26] jya (jya!~jya@mythtv/developer/jya) has quit (Quit: ZNC -
[08:33:47] jya (jya!~jya@mythtv/developer/jya) has joined #mythtv
[08:41:59] Steve-Goodey (Steve-Goodey! has joined #mythtv
[08:42:01] SteveGoodey (SteveGoodey! has quit (Ping timeout: 260 seconds)
[08:49:04] joki (joki! has quit (Ping timeout: 258 seconds)
[08:55:45] joki (joki! has joined #mythtv
[09:00:50] jya: stuartm: the set_pkt is a pain.. I don’t like av_copy_packet as it leaks the data (the existing buffer is simply replaced with the one from the newpkt and the reference count is increased
[09:01:17] jya: AVPacket is now reference counted, but we don’t use it that way
[09:01:59] jya: so we would have to free it, and each time re-create the packet.. otherwise the amount leaked is the same as the amount of data we’ve read. a leak isn’t too much an issue here as mythtranscode doesn’t live very long
[09:02:56] jya: that’s why it is coded as it is now, there’s minimum memory allocation and resizing…
[09:13:59] dekarl1 (dekarl1! has joined #mythtv
[09:16:30] dekarl (dekarl! has quit (Ping timeout: 246 seconds)
[09:39:24] Warped (Warped!~Warped@ has quit (Ping timeout: 250 seconds)
[09:51:13] dekarl1 is now known as dekarl
[09:55:17] gigem (gigem!~david@mythtv/developer/gigem) has quit (Ping timeout: 245 seconds)
[09:59:07] Warped (Warped!~Warped@ has joined #mythtv
[10:29:56] coling (coling! has quit (Ping timeout: 272 seconds)
[10:32:37] Steve_Goodey (Steve_Goodey! has joined #mythtv
[10:33:09] Steve-Goodey (Steve-Goodey! has quit (Ping timeout: 246 seconds)
[10:39:13] Warped (Warped!~Warped@ has quit (Ping timeout: 272 seconds)
[10:49:04] SteveGoodey (SteveGoodey! has joined #mythtv
[10:49:59] Steve_Goodey (Steve_Goodey! has quit (Ping timeout: 272 seconds)
[10:57:12] Warped (Warped!~Warped@ has joined #mythtv
[10:59:33] Steve-Goodey (Steve-Goodey! has joined #mythtv
[11:00:18] stuartm: jya: no longer crashes at that point :) Unfortunately it's still crashing on at least one file here when using --honourcutlist
[11:00:21] SteveGoodey (SteveGoodey! has quit (Ping timeout: 272 seconds)
[11:02:14] stuartm:
[11:05:40] stuartm: we're calling av_free( immediately after set_pkt(), simple double free, will fix
[11:41:52] Steve-Goodey (Steve-Goodey! has quit (Ping timeout: 240 seconds)
[11:42:00] Steve-Goodey (Steve-Goodey! has joined #mythtv
[12:14:42] rsiebert_ (rsiebert_! has joined #mythtv
[12:17:27] rsiebert (rsiebert! has quit (Ping timeout: 246 seconds)
[12:28:06] stuartm: bah, segfault whack a mole, yet another one in FrameInfo
[12:47:32] stuarta: :)
[12:53:17] stuartm: qstring segfault, not exactly sure what's up with it, past experience with those are that 9 times out of 10 they are symptoms of memory corruption but ...
[12:55:56] stuartm: heh, missed the smoking gun – *** Error in `mythtranscode': malloc(): memory corruption: 0x00000000017bd490 ***
[12:56:11] stuarta: oops
[12:59:00] stuartm: that's opened a can of worms :(
[12:59:31] stuarta: why not run it under valgrind, and see what it has to say?
[13:01:18] stuartm: that's my next move, this is just sucking time away from the things I'd rather be doing
[13:01:32] stuartm: c'est la vie
[13:45:32] Warped (Warped!~Warped@ has quit (Ping timeout: 250 seconds)
[13:46:30] stuartm: fix the memory corruption (by letting it leak) and it crashes somewhere else again
[13:46:47] stuartm: this is not fun
[13:48:34] stuarta: sheesh, what is up with that code?
[13:51:38] stuartm: hmm, thinking I didn't fix the corruption after all since the new crash makes no sense at all
[13:52:31] Jordack (Jordack! has joined #mythtv
[13:53:40] stuartm: valgrind reports a bunch of invalid reads and frees, making me think that pkt.size isn't being correctly updated along with somewhere, but that's just short of random guesswork
[13:55:24] stuarta: chuck it in a pastebin?
[13:56:24] stuartm: ok, now we're getting somewhere, an invalid write just after using ensure_size (av_grow_packet), which was causing the crashing last night too ... thinking that grow_packet is either broken or our usage of it is
[14:02:38] dekarl: that's the programmers blues :)
[14:05:00] stuartm: ah-hah, see my first solid bug
[14:05:10] Warped (Warped!~Warped@ has joined #mythtv
[14:05:15] stuartm: and likely cause of some of this corruption
[14:05:51] ** stuarta wonders if \o/ is required **
[14:07:00] stuartm: still segfaults
[14:09:22] stuartm: we were increasing the size of a packet to insert a header in front, memmove was being called to make space for the header but it was using the packet size after the realloc instead of before  – memmove(x+10, x, x+10);
[14:10:01] rmeden (rmeden! has left #mythtv ()
[14:10:17] stuartm: i.e. we'd overrun by 10 in that example
[14:10:18] rmeden (rmeden! has joined #mythtv
[14:11:21] stuartm: yay, well it sorta works, it will run to completion under valgrind
[14:11:42] stuartm: no more invalid write warnings from valgrind, still some invalid reads
[14:11:59] stuartm: but it actually finished transcoding the file
[14:14:12] stuarta: \o/
[14:32:23] gigem (gigem!~david@mythtv/developer/gigem) has joined #mythtv
[14:54:52] stuartm: this is what's left from valgrind –
[14:58:44] stuartm: well that's what's produced per packet – the actual error count is 112075
[15:00:33] stuartm: if I comment out the av_free_packet() in set_pkt then there's no crashing at all
[15:23:08] FabriceMG (FabriceMG! has quit (Quit: FabriceMG)
[16:53:36] Steve-Goodey (Steve-Goodey! has quit (Ping timeout: 250 seconds)
[16:53:42] Steve-Goodey (Steve-Goodey! has joined #mythtv
[17:11:49] Merlin83b (Merlin83b!~Daniel@2a00:1ee0:3:1337:7506:e5c7:b8be:c1cf) has quit (Quit: Leaving)
[17:28:27] andreaz (andreaz! has joined #mythtv
[17:50:13] gary_buhrmaster (gary_buhrmaster!~garybuhrm@2601:9:4980:9d3:222:4dff:fe51:6728) has joined #mythtv
[17:50:33] gary_buhrmaster (gary_buhrmaster!~garybuhrm@2601:9:4980:9d3:222:4dff:fe51:6728) has quit (Remote host closed the connection)
[18:15:23] jya: stuartm: We have to rework the code. Avpacket are now using refcounted memory. But we manipulate the data directly. And not even the right pointer. can't be modified without also modifying where the memory is actually held: in pkt.buf
[18:17:08] jya: av_copy_packet doesn't actually copy the data as such. It simply copy pointers and change the refcount.
[18:20:42] stuartm: jya: I went around fixing all the places that I could find to use the appropriate methods, but still haven't found the one which is causing the segfault
[18:23:54] stuartm: e.g. this was what I had tried by the last build –
[18:24:27] stuartm: it's not using AVPacket properly, but it fixed some of the more problematic bits I found
[18:25:03] stuartm: or in theory it fixed them, but until it stops crashing I don't know if they introduce problems of their own
[18:29:21] Corin_EU (Corin_EU!miller@ has joined #mythtv
[18:30:40] Corin_EU: In the header file include/mythtv/mythversion.h there is the line #include "version.h" . My question is, from which source is it expecting to include the "version.h" file?
[18:31:46] stuartm: jya: it might not be the ideal solution, but those changes get the number of errors reported by valgrind down from 112,075 to 20
[18:32:35] stuartm: Corin_EU: it's generated by configure at compile time
[18:33:59] Corin_EU: So in the same directory as mythversion.h there should be a version.h as well? or in the source which is including mythversion.h  ? The reason I am asking is because I have run into a problem with compiling mythplugins, specifically mytharchive
[18:34:15] Corin_EU: In file included from ../mytharchivehelper/main.cpp:59:0:
[18:34:30] Corin_EU: /usr/local/stow/mythtv-0.27_fixes/X11R6/include/mythtv/mythversion.h:6:21: fatal error: version.h: No such file or directory
[18:35:25] Corin_EU: Hence I am trying to ascertain which version.h it is trying to use and why it is not finding it (eg inlude path problem)
[18:37:20] Steve_Goodey (Steve_Goodey! has joined #mythtv
[18:37:21] Steve-Goodey (Steve-Goodey! has quit (Ping timeout: 244 seconds)
[18:42:15] SteveGoodey (SteveGoodey! has joined #mythtv
[18:42:15] stuartm: Corin_EU: in the root directory
[18:42:41] stuartm: Corin_EU: have you installed mythtv first? Plugins build against the installed sources (normally)
[18:43:19] Corin_EU: YEs mythtv has been configured (with binfdings for perl, php, and python) and installed first.
[18:43:26] stuartm: that's a rather unusual install location – /usr/local/stow/mythtv-0.27_fixes/X11R6/include/
[18:43:42] Corin_EU: Are you familiar with stow?
[18:44:29] stuartm: not at all
[18:44:33] Corin_EU: I know some software developers do not like people to use it for target installations.
[18:45:39] Corin_EU: It uses a very simple method actually. When a sofware installation is to be done, one uses prefix or destdir or whatever to get is installed under /usr/local/stow
[18:45:45] Corin_EU: package_name
[18:45:50] Steve_Goodey (Steve_Goodey! has quit (Ping timeout: 250 seconds)
[18:46:31] Corin_EU: then when everying is installed there, run the stow perl script which creates the links to /usr/local/{bin,lib,include} etc which are located under the package name
[18:46:43] Steve-Goodey (Steve-Goodey! has joined #mythtv
[18:47:08] SteveGoodey (SteveGoodey! has quit (Ping timeout: 250 seconds)
[18:47:27] Corin_EU: I think you get the idea — keep all the installed files under /usr/local/stow/package_name, but make the file available as symbolic links under /usr/local/whatever
[18:48:48] Corin_EU: It really does make management of "local" installed software possible and not have to worry about makefile uninstalls
[18:49:49] Corin_EU: Home page is at
[18:51:08] Corin_EU: What I do not understand is how qmake is getting the include directory under /usr/local/stow/mythtv_0.27_fixes rather than just using /usr/local/X11R6/include
[18:52:04] Corin_EU: ll /usr/local/X11R6/include | grep myth
[18:52:04] Corin_EU: 0 lrwxrwxrwx 1 root sys 49 2014-10–06 17:57 mythtv -> ../../stow/mythtv-0.27_fixes/X11R6/include/mythtv/
[18:52:14] Corin_EU: unless it is being clever and following the symbolic link
[18:53:59] Corin_EU: Hmm, would this be the version.h which it is trying to find? ./libs/libmythbase/version.h because that has not been installed at installation time for some reason
[18:55:58] Corin_EU: all the other version.h files which are generated by a myhtv configure are ./external/FFmpeg/libav*/version.h files
[18:57:43] Corin_EU: stuartm: so is the ./libs/libmythbase/version.h the appropriate version.h file to be included by mythversion.h or not?
[18:58:16] stuartm: yes
[18:59:10] Warped (Warped!~Warped@ has quit (Ping timeout: 272 seconds)
[18:59:52] Corin_EU: I wonder why it has not been installed then
[19:02:19] Corin_EU: I see lots of header files under /usr/local/X11R6/include/mythtv/libmythbase/ but no version.h
[19:06:09] Corin_EU: I do not see any mention of version.h to be installed in mythtv-0.27_fixes/libs/libmythbase/Makefile
[19:07:47] Corin_EU: -$(INSTALL_FILE) /usr/src/build/mythtv-0.27_fixes/libs/libmythbase/mythdbcon.h $(INSTALL_ROOT)/usr/local/X11R6/include/mythtv etc etc etc
[19:07:52] Corin_EU: but not for version.h
[19:08:20] Corin_EU: So is this a bug?
[19:09:00] Corin_EU: -$(INSTALL_FILE) /usr/src/build/mythtv-0.27_fixes/libs/libmythbase/mythversion.h $(INSTALL_ROOT)/usr/local/X11R6/include/mythtv is there
[19:11:00] Corin_EU: but no install line for just version.h
[19:17:10] Warped (Warped!~Warped@ has joined #mythtv
[19:20:48] stuartm: jya: got it down to zero errors from valgrind and no crashing, still not quite full-rewrite you proposed but I'd call it progress
[19:21:39] stuartm: however those last twenty errors required making a second copy in ConvertToI
[19:22:01] stuartm: I think now I understand where it's going wrong that I may be able to avoid that
[19:26:23] coling (coling! has joined #mythtv
[19:26:53] Corin_EU: stuartm: I see inc.files += compat.h mythversion.h mythconfig.h mythconfig.mak version.h in the file
[19:29:12] stuartm: Corin_EU: could be a newly introduced bug in the order we create version.h and when qmake processes the project files, if qmake checks that the file exists before building the makefile then it may just ignore it
[19:35:22] Warped (Warped!~Warped@ has quit (Ping timeout: 240 seconds)
[19:38:04] Corin_EU: The only thing is that the line in the project file with version.h has five entries and version.h is the last, whereas the other lines have four or less. I am just in the proccess of checking whether or not that is relevant by putting version.h in a different position
[19:39:14] Corin_EU: Incidentally after manually copying that version.h to the mythtv include directory, I am most relieved to see that mythplugins build successfully!
[19:42:18] stuartm: Corin_EU: the number of entries per line makes no difference, it's probably what I suggested but I'll need to check
[19:44:41] Corin_EU: if qmake checks that the file exists before building the makefile — you mean that because version.h does not exist when qmake creates the makefile, it does not put it in as an install line? That seems bizarre
[19:47:08] stuartm: it's the only possible explanation I can think of
[19:48:00] stuartm: that doesn't mean it's the right one, but it's worth excluding that possibility first
[19:51:12] Corin_EU: Okay. but I am stepping through the build, and just noticed that version.h is created before the Makefile
[19:51:31] Corin_EU: 8 -rw-r--r-- 1 root root 4985 2014-10–06 20:42
[19:51:31] Corin_EU: 52 -rw-r--r-- 1 root root 49220 2014-10–06 20:46 mythconfig.h
[19:51:31] Corin_EU: 48 -rw-r--r-- 1 root root 45260 2014-10–06 20:46 mythconfig.mak
[19:51:31] Corin_EU: 4 -rw-r--r-- 1 root root 114 2014-10–06 20:50 version.h
[19:51:37] Corin_EU: no makefile created yet
[19:52:43] Corin_EU: Doing the ffmpeg stuff now, so it will take a while to get the libmythbase ...
[19:52:46] stuartm: came to the same conclusion looking at mythtv/Makefile
[19:52:56] Corin_EU: :+)
[19:55:11] Warped (Warped!~Warped@ has joined #mythtv
[19:56:07] Corin_EU: I can just add a custom "copy version.h" to my build specification file for the time being to get around the problem.
[19:57:08] Corin_EU: As I originally explained, my difficulty was knowing which version.h it should be  ;+) So thanks for all of your time and patience in explaining.
[20:01:49] andreaz (andreaz! has quit (Quit: Leaving.)
[20:02:52] andreaz (andreaz! has joined #mythtv
[20:06:56] andreaz (andreaz! has quit (Client Quit)
[20:07:08] andreaz (andreaz! has joined #mythtv
[20:14:01] Warped (Warped!~Warped@ has quit (Ping timeout: 260 seconds)
[20:23:07] coling (coling! has quit (Ping timeout: 245 seconds)
[20:31:52] skd5aner (skd5aner! has quit (Read error: Connection reset by peer)
[20:32:18] skd5aner (skd5aner! has joined #mythtv
[20:34:48] dekarl (dekarl! has quit (Ping timeout: 246 seconds)
[20:44:27] Cougar (Cougar!~cougar@2a03:5880:104:10:5579:413:8f11:91cd) has left #mythtv ("Client exiting")
[20:50:21] coling (coling! has joined #mythtv
[20:51:36] Steve-Goodey (Steve-Goodey! has quit (Quit: Konversation terminated!)
[21:22:42] Warped (Warped!~Warped@ has joined #mythtv
[21:34:27] Warped (Warped!~Warped@ has quit (Ping timeout: 245 seconds)
[21:41:37] Jordack (Jordack! has quit ()
[21:46:46] andreaz (andreaz! has quit (Read error: Connection reset by peer)
[21:53:25] paul-h (paul-h!~Paul@ has joined #mythtv
[21:54:45] paul-h: jya: Is #10919 still a problem it's an old ticket and a lot has changed?
[21:54:45] ** MythLogBot **
[21:57:55] paul-h (paul-h!~Paul@ has quit (Remote host closed the connection)
[21:58:59] paul-h (paul-h!~Paul@ has joined #mythtv
[22:02:03] Warped (Warped!~Warped@ has joined #mythtv
[22:07:45] paul-h: Anybody know if Hauppage HVR-1950 is working with 0.27? #10940
[22:07:45] ** MythLogBot **
[22:21:08] Tobbe5178 (Tobbe5178! has quit (Read error: Connection reset by peer)
[22:24:21] paul-h (paul-h!~Paul@ has quit (Quit: Konversation terminated!)
[22:34:56] gigem: FYI, I just pushed a devel/scheduler branch. This branch has the first public version of the affinity changes I've hinted at. I'm now looking for Guinea Pigs, I mean testers for it. :) I'm most interested in hearing from people who make heavy use of multirec. At this point, I'm not expecting anyone to run it on a production system. I just want to see --testsched output from this version and production to
[22:34:58] gigem: compare. I can make a patch against 0.27-fixes. In fact, I will probably patch my production version of 0.27 to test on.
[23:21:48] Corin_EU (Corin_EU!miller@ has quit (Quit: Leaving)
[23:25:30] jya: stuartm: are all your changes to mythtranscode committed ?
[23:25:48] jya: much smaller change than I expected
[23:26:12] stuartm: jya: no, I've attached a patch to ticket #12280
[23:26:12] ** MythLogBot **
[23:27:13] stuartm: it works, but that doesn't mean it's 100% right, so I'd like to re-visit it tomorrow to double check and hopefully by then I'll also have some feedback
[23:27:46] jya: stuartm: seeing that we aren’t using AVPacket properly regardless, maybe we can simply continue to do so and go back to . . . b01d8/mythtv
[23:27:58] jya: and don’t use AVPacket at all to prevent confusion
[23:28:00] stuartm: as I noted before, it's not the re-write you suggested and which I agree is needed
[23:31:36] stuartm: jya: we'll see how the patch goes, I'm not sure how far we'd have to go back to find a time when we didn't use AVPacket at all
[23:32:10] jya: no no.. what I mean is we can use our own memory buffer, call that structure whatever you want..
[23:32:19] jya: keep the code almost identical to what it’s always been
[23:32:34] jya: but only create an AVPacket when we need to pass it to ffmpeg
[23:32:39] jya: from that buffer
[23:33:49] stuartm: ah, well that might be the practical way forward given we mostly use AVPacket as a dumb buffer and not much else, no side data etc
[23:33:58] jya: exactly...
[23:34:19] jya: I don’t even believe we use the pts/dts information
[23:34:53] jya: that modification would be much simpler to code, and also much easier to maintain going forward as it won’t depend on AVPacket and how ffmpeg decide to change that structure
[23:35:58] stuartm: jya: we do use the pts/dts, lossless transcode has to renumber the frames/adjust timestamps when cutting out adverts etc
[23:36:24] stuartm: but it's still no more than a simple struct (iirc)
[23:38:20] jya: looking at your change, that code must have been leaking a long time !
[23:38:31] jya: especially the pkt = spare->pkt; line 1784
[23:40:39] stuartm: yeah, some of the issues seem to have been long standing, e.g. even though the memory corruption fix which I pushed is relatively new, it's only due to some bad code that it ever worked previously
[23:42:38] stuartm: when we realloc'd the buffer in older code we simply failed to increase pkt.size ... so although that prevented us using a memcpy with an incorrect length, it probably had knock on effects elsewhere
[23:44:56] stuartm: I was only just starting to piece everything together when I fixed the last valgrind issue (and the segfault), I don't know if I want to wade any deeper to find the issues which aren't yet causing problems for most people
[23:56:10] gregL (gregL! has quit (Quit: Leaving)
[23:56:33] stuartm: might make it a project for 0.29 so I can focus on the stuff I want in 0.28, the sooner that's done, the earlier 0.28 will be released

IRC Logs collected by BeirdoBot.
Please use the above link to report any bugs.