Trac:
Patch ticket: http://bugs.bitlbee.org/bitlbee/ticket/1098
OpenBSD ticket: http://bugs.bitlbee.org/bitlbee/ticket/1003

It crashes the damn thing. Their bug but also our bug.

Trac: http://bugs.bitlbee.org/bitlbee/ticket/917#comment:6

This Robert had a backtrace but it got lost in an expired pastebin. It was something along the lines of account_on adding contacts right away and dereferencing some null pointer in the whatever structs of whatever whatsapp-purple things. That.

References #19

libcurl

< jgeboski> dx: another thing that should be done if the API is going to break... fix the damn pkgconfig include dir
< jgeboski> to be bitlbe/*.h
< jgeboski> for instance, global includes for url.h, it's just #include <url.h>
< jgeboski> not bitlbee-url.h
< jgeboski> or bitlbee/url.h
< jgeboski> it's very uncleanly
< jgeboski> and could very likely lead to header conflicts

< jgeboski> you can see a ton of things that are ambiguously named
< jgeboski> and having them in a global include path could be severely problematic
< jgeboski> the only reason this has not been an issue as of yet is because there are not many plugins, and most other libraries do things correctly in this department

< dx> so this is just removing -I/usr/include/bitlbee from bitlbee.pc and changing includes to #include <bitlbee/url.h>?
< jgeboski> yes
< jgeboski> it might require some changes in the configure/makefiles for bitlbee, but all of those changes should be trivial as well
< dx> neat

http://bugs.bitlbee.org/bitlbee/ticket/1114

srsly dude

the plan is to send IDs coded with a timestamp in them, and if the error has a valid timestamp, show formatted time in the ui error message.

This one is for @jcopenha, who cares about the portability of his resolv patch much more than i do.

So we have this:

if detect_resolv_dynamic || detect_resolv_static; then
    echo '#define HAVE_RESOLV_A' >> config.h
    if detect_resolv_ns_dynamic || detect_resolv_ns_static; then
        echo '#define HAVE_RESOLV_A_WITH_NS' >> config.h
    fi
else

And this debian bug: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=291609

Which causes the detect_resolv_ns_dynamic check to fail with:

/tmp/ccyZuKqf.o: In function `main':
:(.text+0x18): undefined reference to `__ns_initparse'
:(.text+0x31): undefined reference to `__ns_parserr'
collect2: ld returned 1 exit status

And pretty much the same thing at the end of the build:

lib/lib.o: In function `srv_lookup':
ns_parse.c:(.text+0x4e44): undefined reference to `__ns_initparse'
ns_parse.c:(.text+0x4f60): undefined reference to `__ns_parserr'

In this particular failing case (with emoji because why not)

• ✅ detect_resolv_dynamic succeeds
• ⚠️ detect_resolv_static isn't checked
• ❌ detect_resolv_ns_dynamic fails (debian bug)
• ✅ detect_resolv_ns_static succeeds

Since detect_resolv_dynamic / detect_resolv_static are in charge of adding themselves to EFLAGS, this ends up trying to link against the libresolv.so, while needing libresolv.a for resolv_ns.

Soooo, wtf do we do?

seirl is doing this

from @jgeboski's suggestions - split to individual tickets as needed.

General:

• bump minimum glib version requirement to 2.16 minimum, 2.24 maybe later
• remove backwards compat crap (TODO: what crap?)
• use g-types instead of checking for c99 stuff ourselves

Specific:

• use glib's ini parser for bitlbee.conf (glib 2.6 mostly)
• use glib's base64 encoder (glib 2.12)
• use glib's md5 (glib 2.16)
• use glib's sha1 (glib 2.16)
• g_strcmp0() (glib 2.16)
• GHashTableIter (glib 2.16)
• GIO's GResolver (glib 2.22)
• g_unix_signal_add() (glib 2.30) solved it with pipes
• g_slist_free_full() (glib 2.28)

Rejected:

• nuke event_libevent and b_io_* wrappers, use glib directly everywhere?

• Rebase branches and merge all the pending patches
  • for-wilmer-19-jan
  • wip/md5
  • feat/hip-cat (welp)
  • patch/twitter-filter (jgeboski)
  • patch/mode-special (jgeboski)
  • wip/jabber-subprotocols
  • patch/headers
  • patch/use-dev-urandom-properly
  • flexo/python
  • WPettersson:twitter_format_strings
  • WPettersson:rawreply
  • WPettersson:threads (he will take care)
  • patch/carbons
• Reindent (5983eca)
• Some nightly build system
• Poke everyone who starred or watched this repo to move over

http://bugs.bitlbee.org/bitlbee/query?status=!closed&keywords=~blist

The fix for the missing arpa/nameser.h types on OpenBSD broke the cygwin build. The detect_nameser_ns_types code is not called and does not correctly compile on other platforms causing redundant type definitions.

Importing bugs I care about from trac.

http://bugs.bitlbee.org/bitlbee/ticket/271

⚠️

i mean why the hell not

Trac: http://bugs.bitlbee.org/bitlbee/ticket/1085

Don't remember what was missing to apply this one...

Fuck yeah coverity

76c85b4#diff-b6da343f1913e438b4650d5b04bfcfcaR388

new = new->next, new becomes new->next, old becomes forgotten, new is returned

parson

this is going to hurt

DO SOMETHING

Old "patch": http://bugs.bitlbee.org/bitlbee/ticket/971

Arch's version of tmpfiles: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/bitlbee.tmpfiles?h=packages/bitlbee

Don't think we can predict the location that needs to be cleared:

pidfile='/var/run/bitlbee.pid'
ipcsocket=''

Also the ipc socket gets disabled with this, can't be re-enabled in runtime

adding tickets to track stuff i want to get done before the next release

this would be loosely based on the changes the carbons patch does.

Channels that show debug.

Importing bugs I care about from trac.

http://bugs.bitlbee.org/bitlbee/ticket/1114

Trac: http://bugs.bitlbee.org/bitlbee/ticket/785

There's a patch in there, but it's not correct.

I've followed this wiki page: http://wiki.bitlbee.org/HowtoHipchat

But I get Unknown protocol when trying to
account add hipchat [email protected] <password>

Is there a step missing in the wiki page? Do I need to change some settings in configure when making bitlbee?

Importing bugs I care about from trac.

http://bugs.bitlbee.org/bitlbee/ticket/1122

adding tickets to track stuff i want to get done before the next release

BASH PROGRAMMING TOO COMPLEX

GET HYPE

Trac: http://bugs.bitlbee.org/bitlbee/ticket/1044

Half a patch written for this one. Should spend more than 5 minutes on it.

Our test suite is not great but the github integration is so sweet~