Hi,

Here's the summary of the previous IRC meeting / sprint.

---

COMMUNITY MEETING

Place: #openvpn-devel on irc.freenode.net
List-Post: openvpn-devel@lists.sourceforge.net
Date: Thursday 24th Nov 2011
Time: 18:00 UTC

Planned meeting topics for this meeting were on this page:

<https://community.openvpn.net/openvpn/wiki/Topics-2011-11-24>

Next meeting will be announced in advance, but will probably be on the same
weekday and at the same time. Your local meeting time is easy to check
from services such as

<http://www.timeanddate.com/worldclock>

or with

$ date -u


SUMMARY

andj, cron2, dazo, krzie and mattock participated in this meeting.

--

Discussed current state of OpenVPN buildsystems. Agreed that for 2.2.2
and 2.3 releases no big changes should be made. However, rethinking the
(Windows) building after 2.3 would make sense: for example, split the
TAP-driver into it's own (sub)project and build and sign the OpenVPN
binary for Windows on Cygwin + mingw_w64. For details, see

<https://community.openvpn.net/openvpn/ticket/174>

--

Discussed "IPV6_RECVPKTINFO vs. IPV6_PKTINFO" patch briefly:

<http://thread.gmane.org/gmane.network.openvpn.devel/4775>

This patch has not yet been ACKed. Mattock will ask JJO to review this.

--

Discussed the "openvpn-2.2.1-install.exe contains unsigned
openvpn-gui-1.0.3.exe" bug:

<https://community.openvpn.net/openvpn/ticket/171>

Mattock will fix this for 2.2.2.

--

Attempted to generate an OpenVPN installer from "master" that works for
all supported Windows versions in both IPv4 and IPv6 modes. The
following installer seems to work ok on WinXP 32-bit and Win7 32-bit:

<http://build.openvpn.net/downloads/snapshots/openvpn-2.3-openvpn-inet-ntop-install.exe>

The following two bugs have been fixed in this installer:

<https://community.openvpn.net/openvpn/ticket/126>
<https://community.openvpn.net/openvpn/ticket/169>

For some reason the installer did not work on mattock's Windows 7 64-bit
VM. If you want to try this out on that platform, you need to put
Windows 7 64-bit into test mode. For details, look here:

<https://community.openvpn.net/openvpn/wiki/BuildingOnWindows#Method2:Enablingtestmodeontargetcomputer>

Also note that WinXP might not have IPv6 support installed by default.

--

Discussed OpenVPN 2.3 release in general. Current ticket list can be
viewed from here:

<https://community.openvpn.net/openvpn/report/3>

Agreed that releasing the first 2.3 alpha in January is doable.

--

Discussed various bugs related to <connection> profile handling:

<https://community.openvpn.net/openvpn/ticket/16>
<https://community.openvpn.net/openvpn/ticket/19>
<https://community.openvpn.net/openvpn/ticket/78>

Decided to postpone fixing profile handling for good until 2.4, so that
2.3 release does not get delayed any further.

--

Discussed the "OpenVPN Install and the Windows Path Environment
variable" bug:

<https://community.openvpn.net/openvpn/ticket/34>

Mattock will do some research to find out how difficult fixing this
would be.

--

Discussed the "Option --register-dns does not work in conjunction with
--win-sys" bug:

<https://community.openvpn.net/openvpn/ticket/106>

There are still some hardcoded paths in win32.c in latest "master", one
of them in a critical place.

---

Full chatlog as an attachment

-- 
Samuli Seppänen
Community Manager
OpenVPN Technologies, Inc

irc freenode net: mattock

mattock 19:58:42
almost meeting time...  

andj 19:59:17
I'm going to miss most of the meeting I'm afraid        

cron2 20:01:48
so      
cron2 is here! 20:01    

mattock 20:02:31
will dazo attend today? 

cron2 20:05:11
haven't seen anything   

mattock 20:05:34
well, I'll check my own list of "things to do before 2.2.1"
...while waiting 20:05:43
 
cron2 20:05:50
we're at "before 2.2.2" now     

mattock 20:06:41
oops    

cron2 20:07:30
let me briefly test the openvpn.exe I built this morning, so I can say for sure 
whether my patching worked      

mattock 20:07:43
ok
this bug report is still unresolved: 
https://community.openvpn.net/openvpn/ticket/126 20:07:53
then there's the "IPV6_RECVPKTINFO vs. IPV6_PKTINFO" patch, which was sent to 
jjo for review 20:08:21
no clue what it's current status is 20:08:27
 
cron2 20:08:49
#126 is unresolved because we don't know whether the 9.9 tap driver works on 
Win7
it works for me on XP and fixes the small-packet issue 20:09:02
mattock: I think we need a signed tap driver for the reporter of #126 to test 
20:11:17
 
mattock 20:11:29
ok, it seems IPV6_RECVPKTINFO patch has not gone anywhere... from jjo on 10th 
Aug: "Thanks a lot for looping me in, afaics at a glance I'd propose a bit
diff approach, to avoid "generalizing" the osx mishandling of these symbols 
into other platforms. Will take a look on the weekend && reply."
maybe I'll push jjo a little more 20:11:41
 
cron2 20:11:47
mattock: so if you could get James to sign the driver, this could be tested and 
closed for good 

mattock 20:12:14
so this #126 is what your recent patches fix?   

cron2 20:12:21
yes     

mattock 20:12:24
ah, ok
I'll mail james 20:12:31
 
cron2 20:12:39
the patches to tapdrvr.c, and for completeness, version.m4 and tun.c    

mattock 20:12:41
maybe he could provide signed 9.9 tap drivers   

cron2 20:12:52
that would be great, so win7 people could then test     

mattock 20:14:59
mail sent       

cron2 20:16:55
*sigh*, windows is hating me again      

mattock 20:18:49
updated #126
...now 20:19:01
 
cron2 20:20:18
mmmh, the binary I built today is not working at all, but maybe I forgot some 
sort of finalize step
mattock: could you build an installer with "master" plus 
http://pastebin.com/Zr7T9jrv 20:21:58
so we can see whether that one works on XP now? 20:22:12
I think it should 20:22:15
 
mattock 20:22:16
ok      

cron2 20:23:00
most likely I'm just not supposed to copy-away the openvpn.exe from the build 
directory before some manifest thingie is updated, or so  

mattock 20:27:23
patching
argh, issues with the patch format 20:31:35
 
cron2 20:31:59
huh, what is it disliking?      

mattock 20:32:51
not sure, it can't detect the patch format
I'll try the downloaded version with plain patch 20:33:08
http://pastebin.com/sGbJ4xdr 20:34:04
it apparently does not like the git line 20:34:15
 
cron2 20:34:32
just throw it out, then 
and the "index" line as well 20:35:02
 
mattock 20:36:27
did not help, got tons of rejects       

cron2 20:36:46
I'll put up a proper git format patch. mom.     

mattock 20:36:54

I was about to ask 20:36:57
 
cron2 20:37:22
no idea what happened, must have been windows mangling things   

mattock 20:37:31
might be, it was in DOS format  

cron2 20:40:55
mattock: you have mail  

mattock 20:41:19
not yet, but probably soon 
arrived 20:42:59
applied cleanly 20:43:33
damn, those inet_ntop errors (yet) again 20:45:07
 
cron2 20:45:16
wut?
oh 20:45:25
 
mattock 20:45:27
I'll double check and then paste an error log   

cron2 20:45:29
patch is incomplete     

mattock 20:45:31
ok      

cron2 20:46:17
please change the two lines above the #define inet_ntop() in win32.h to read 
"int openvpn_inet_pton..." and "const char * openvpn_inet_pton". Missed that 
part when re-doing the fix. Or just move the two #define lines two lines up, 
before the "inet_ntop" line
gah 20:46:22
 
mattock 20:46:59
just sec
a 20:47:01
 
cron2 20:48:45
mattock: will send you a new patch      

mattock 20:49:01
just patched it myself, but feel free   

cron2 20:49:31
sent    

mattock 20:49:58
looks better now
built and packaged 20:51:41
 
cron2 20:51:47
good    

mattock 20:51:50
I'll copy the installer + tap-driver + openvpn.exe to build.openvpn.net 

cron2 20:52:17
I'll download and test the installer then       

mattock 20:55:58
ok, here: http://build.openvpn.net/downloads/snapshots/ 

cron2 20:56:04
yeah, downloading       

mattock 20:56:04
it's the inet-ntop stuff        

cron2 20:59:05
meh     

mattock 20:59:50
I wonder if I have to install another Win7 64-bit test VM to get rid of the 
messy driver situation      

cron2 20:59:53
the binary starts, but then IPv6 fails
the replacement functions do not work properly 21:00:28
I use inet_pton() to ask the system "is this a valid IPv6 address?" and I get 
back "no!" for everything 21:01:35
 
mattock 21:03:07
time for a custom IPv6 address verifying function using regular expressions?    

cron2 21:03:15
nah
if inet_pton() isn't working, we get more problems in other places 21:03:27
 
mattock 21:04:46
cron2: do you want to take a stab at fixing this today?
it seems the official meeting is stub today 21:04:55
 
cron2 21:07:23
mattock: well, as it seems nobody else is here, let's briefly discuss what we 
have on the agenda        

mattock 21:07:31
ok
build system? 21:07:38
I was wondering if Cygwin could be used for (official) Windows builds 21:08:06
except for TAP-driver building + signing 21:08:13
 
cron2 21:08:40
you can build the TAP driver on mingw+msys, so I'm sure you can build it on 
cygwin+mingw64 - you just need the windows ddk/sdk installed
so it's really up to you (and James) to decide what the official build system 
should be like - you're the ones that have the nice task of building "official" 
installers... 21:09:13
 
mattock 21:09:20
currently windows building is a mess, and visual studio is constantly causing 
issues    
cron2 will try to make MSVC builds work, but prefers (by FAR) anything that is 
gcc-based 21:09  

mattock 21:10:21
the VS/Python builds works really well now, but having to use the VS C compiler 
seems to be problematic
when coding, that is 21:10:32
or that's my feeling 21:10:36
 
cron2 21:10:59
it's very strict about things, which is problematic when trying to do portable 
stuff    

mattock 21:11:14
I think we should stick with VS/Python for official builds for 2.3 in any case
hmm, I wonder if it could be made less strict somehow 21:11:22
I think I'll try Cygwin builds in the near future to get a feel for them 
21:13:37
 
cron2 21:13:46
I don't think we can decide much here today - needs input from James, testing 
Cygwin, etc.      

mattock 21:13:49
yep     

cron2 21:13:58
2.2.2 release   

mattock 21:14:06
ok, so here's the list of tickets per milestone: 
https://community.openvpn.net/openvpn/report/3 

cron2 21:14:09
what we need is a signed + tested tap 9.9 driver        

mattock 21:14:23
yep
and I probably need to reinstall Win7, as the tap-driver behavior is not too 
encouraging 21:14:41
 
cron2 21:14:59
ouch, that is a lot of open bugs
I think we should focus on "tap driver fix", "get xp to work again" and "remove 
the warning in #147" 21:15:45
 
mattock 21:16:33
yeah
some of the bugs are "legacy" (from sf.net bug tracker) 21:16:44
although many were removed before moving them over, and even after that 21:16:57
this we can probably manage with a small effort: 
https://community.openvpn.net/openvpn/ticket/127 21:17:34
 
cron2 21:18:09
yeah    

mattock 21:18:17
for this we need to poke the reporter: 
https://community.openvpn.net/openvpn/ticket/154
I'll do it right now 21:18:32
ah, dazo should get here 21:20:08
he said he'll be "20-30 minutes late" 21:20:18
so, I guess he forgot to check "date -u" 21:20:31
 
cron2 21:20:54
patch in your mailbox
applies on top of "master" (includes previous patch) 21:21:05
L'utente dazo_afk è ora conosciuto come dazo 21:21     
dazo here now 21:22     

mattock 21:22:33
hi dazo!        

dazo 21:22:34
sorry about the delay   
dazo catches up on scrollback 21:22     

mattock 21:23:11
I was afraid the meeting would be a stub
I mailed james at around 20:20, so far no response 21:23:21
 
cron2 21:24:23
mattock: could you build a new installer, please?       

dazo 21:24:52
regarding build system core for Windows ... I have no problems with moving 
towards cygwin/msys/mingw/what-ever-its-called .... I see that the TAP driver 
might need VS, but that's a different issue    

cron2 21:25:13
no, not VS, Windows SDK, which is a separate thing from VS      
cron2 has succeeded in building a tap driver with mingw+msys plus wdk 
(domake-win) 21:25        

mattock 21:25:37
james is on a vacation  

dazo 21:25:40
ahh!    

cron2 21:25:48
mattock: when will he be back?  

mattock 21:25:51
good for him, afaik he's been working way too much
not sure, asked him via email 21:25:55
 
dazo 21:26:04
okay, then it's even easier ... I thought VS was tightly hooked into these 
things
To be honest, I'd really like to get the Windows TAP driver out of the stock 
OpenVPN stuff ... it's not related to OpenVPN (except it provides a tun/tap 
device) ... It's the installers task to pull in the needed sub-parts for the 
overall user experience, in my eyes ... so I'm along with Alon's thoughts here 
21:26:42
 
mattock 21:26:44
cron2: the "patch 4" is borked or something     

cron2 21:26:53
mattock: what happens?
dazo: yeah 21:27:01
 
dazo 21:27:12
get a OpenVPN MSI for only OpenVPN stuff ... and WinTAP MSI for the TUN/TAP 
driver ... and the installer ships them both        
cron2 is with dazo on that. for 2.3 or for 2.4? 21:27   

cron2 21:27:38
for 2.2.2 we need to stick to what we have      

dazo 21:27:42
agreed
I'd like to see this already for 2.3 21:27:49
 
mattock 21:27:53
cron2: patch format detection fails (git am), hunks get rejected (patch)        

dazo 21:28:07
Not necessarily the first alpha/beta stuff ... but at least for RCs     

cron2 21:28:10
meh, this is straight out of "git diff" 

dazo 21:28:23
git apply       

cron2 21:28:32
mattock: please do git apply    

dazo 21:28:40
(or patch -p1 if 'git apply' gets grumpy)       

cron2 21:29:03
yeah, git am wants mail headers on top, which are not there     

dazo 21:29:14
yeah    

mattock 21:30:17
git apply failed, patch -p1 worked
interesting how difficult patching can be 21:30:23
 
dazo 21:31:29
hehe ... git is hyper-sensitive to patches .... which I consider a good thing   

mattock 21:32:50
building...
ok, on build now 21:35:42
http://build.openvpn.net/downloads/snapshots 21:35:57
*-patch4-* 21:36:07
 
cron2 21:36:19
installing...
meh, still fails, with WSAEINVAL 21:37:56
 
mattock 21:38:55
this is interesting: https://community.openvpn.net/openvpn/ticket/171   

cron2 21:39:07
mattock: patch5 on its way to you       

mattock 21:39:08
did not realize somebody (James?) had signed the GUI also
cron2: ok 21:39:14
 
cron2 21:39:38
indeed, having a signed gui is useful   

mattock 21:40:16
I'll fix that, it's trivial
patching 21:41:20
 
cron2 21:42:14
oh
gaaah 21:42:21
Support for IPv6 addresses using the WSAStringToAddress function was added on 
Windows XP with Service Pack 1 (SP1) and later. IPv6 must also be installed on 
the local computer for the WSAStringToAddress function to support IPv6 
addresses. 21:42:42
 
mattock 21:42:52
building        

dazo 21:43:19
whoops
cron2: does that mean that you need to install some extra IPv6 patches for the 
new TAP driver to work in XP? 21:43:54
 
cron2 21:43:57
I think my IPv6 failures are just due to "no IPv6 installed on that XP VM"
just enable it 21:44:00
netsh interface ipv6 install 21:44:04
let me re-test 21:44:08
 
mattock 21:44:22
ok
let me know if you need the patch5-based installer 21:44:38
meanwhile I'll skim through all bug reports 21:44:50
 
cron2 21:45:10
works perfectly 
patch4 based, that is 21:45:16
I'll go back to the last one 21:45:21
inet-ntop 21:45:33
that was purely a problem with the XP VM I cloned monday, which - purposely - 
did not have v6 yet 21:46:03
 
dazo 21:46:19
just a side related thought .... WinXP has end of support from Microsoft April 
8, 2014 ... shall we kill XP support at that time as well? ... will that make 
things easier for us?      

mattock 21:46:30
this might be "works for me": https://community.openvpn.net/openvpn/ticket/68   

cron2 21:46:42
dazo: let's keep it for 2.3, and then kill it   

mattock 21:46:43
a custom windows build with an issue    
andj is back 21:46      

mattock 21:47:00
hi andj!        

cron2 21:47:11
mattock: inet-ntop installer works with ipv6 on xp \o/  

mattock 21:47:17
nice!   

cron2 21:47:26
also works with small packets   

dazo 21:47:32
cron2: fair enough ... as long as building XP binaries doesn't get too 
complicated .... like what we experienced with TAP driver for Win2k      

cron2 21:47:34
(ping -l 1)     

dazo 21:48:05
cron2: so, IPv6 needs to be enabled for the new TAP driver to function?
(the netsh stuff) 21:48:14
 
cron2 21:48:15
dazo: no        

andj 21:48:17
how's the meeting going? and any news on blockers for the 2.3 alpha?    

cron2 21:48:28
that's unrelated to the tap driver, that's needed to make inet_pton() accept 
IPv6 addresses
dazo: we can build XP-compatible binaries now 21:48:43
 
dazo 21:48:44
andj: I'm late in as well, but I think it's 2.2.2 stuff which is being solved 
yet 
ack! 21:48:49
 
cron2 21:49:01
dazo: actually it's fixing #169 

mattock 21:49:10
it'd be great if the installer would work on 64-bit Win7, too 
nobody around with test-mode win7 64-bit around? 21:49:27
oops 21:49:30
 
andj 21:50:00
mattock: I have one at work, but not here I'm afraid    

mattock 21:50:32
andj: could you smoketest this installer there: 
http://build.openvpn.net/downloads/snapshots/openvpn-2.3-openvpn-inet-ntop-install.exe
  

andj 21:51:00
yes, but that'll be somewhere tomorrow  

mattock 21:51:01
my win7 seems to reject it and I'm not sure if it's because windows is borked 
or not
that's fine 21:51:08
 
cron2 21:51:26
patch (0001-work-around-inet_ntop-inet_pton-problems-for-MSVC-bu.patch  

mattock 21:51:27
I'd rather not reinstall my win7 VM just to verify that 

cron2 21:51:29
) sent to list  

andj 21:52:48
mattock: I'll see what I can do, I've got a VM image with a win 2008R2 in test 
mode lying around somewhere      

mattock 21:52:57
andj: that'd be great!  

andj 21:53:14
so it isn't win7 as promised earlier    

cron2 21:54:53
so - coming back to 2.2.2 release
do we know when James will come back? (Or is there anyone else who can sign 
drivers)? 21:55:15
do I need a signed driver on win7/32? 21:55:38
 
mattock 21:56:17
cron2: he hasn't responded yet, so no
regarding "anybody else"... no, not atm 21:56:29
for 32-bit Windows signing is not mandatory 21:56:46
 
cron2 21:56:53
ok, let me try win7/32  

dazo 21:56:58
cron2: looking at the patch you sent ... just wondering, you #define inet_* -> 
openvpn_inet_* .... but you declare inet_* functions in socket.c ... shouldn't 
they be openvpn_inet_*?   

cron2 21:57:24
they get renamed on declaration
macros also apply there 21:57:45
if you think this is not clean, I'll change that as well (it was that way 
before I put my hands on it, so I didn't want to change too much) 21:58:14
 
dazo 21:59:07
well, the windows code is pretty nasty spaghetti for me ... so I might lack the 
overall understanding here now  

cron2 21:59:46
yep, tap 9.8 fails on "ping -l 1 -4 ..."
let's try tap 9.9 21:59:49
 
dazo 22:00:28
because I don't see where the openvpn_inet_ntop() function is ... I see it is 
declared in win32.h but not where the "contents" of that function is ...  

cron2 22:00:40
socket.c
search for "just" inet_ntop() and inet_pton() 22:00:51
but since the declaration is after the #define, the function gets renamed right 
away 22:01:02
 
dazo 22:01:42
there is$ git grep openvpn_inet_ntop | wc -l
0 22:01:43
 
cron2 22:01:50
21:00 <@cron2> search for "just" inet_ntop() and inet_pton()
21:00 <@cron2> search for "just" inet_ntop() and inet_pton() 22:01:51
21:00 <@cron2> search for "just" inet_ntop() and inet_pton() 22:01:51
 
dazo 22:01:52
(this is before this patch)     

cron2 22:01:55
21:00 <@cron2> search for "just" inet_ntop() and inet_pton()
socket.c 22:02:02
 
dazo 22:02:36
yeah ... but this define then looks strange: #define inet_ntop(af,src,dst,size) 
openvpn_inet_ntop(af,src,dst,size)
as it defines something known to something unknown 22:02:45
cron2 gives dazo a mug of coffee, and asks him to re-read what I wrote 22:03    

cron2 22:04:55
ok, topic switch: tap driver 9.9 works for me on win7/32, and "ping -l 1 -4 
..." is fixed, so this is fixing the small-ipv4-packets problem for good
topic switch back 22:05:00
dazo: if you think it's more clean, we can rename the inet_ntop and inet_pton 
functions in socket.c to openvpn_inet_ntop and openvpn_inet_pton just fine. I 
just didn't do that (yet) since the #define will take care of this anyway. 
22:05:40
 
mattock 22:06:13
that would probably reduce some confusion       

dazo 22:06:15
cron2: I'm just quite tired these days, so I'm trying to see what you see ... 
but I don't see it yet ... I'm applying the patch to see how things looks like 
then       

cron2 22:06:37
dazo: socket.c has a function "inet_ntop()"
it includes something that includes win32.h 22:06:47
 
dazo 22:06:57
cron2: it might be cleaner to avoid name confusion like this ... by prefixing 
our own "versions" of it  

cron2 22:07:07
so when the compiler reads socket.c, the #define from win32.h is active, and 
the preprocessor will change it to openvpn_inet_ntop()
ok 22:07:12
cron2 goes re-spin the patch 22:07      

cron2 22:11:05
sent    
L'utente uuuppz è entrato nella stanza 22:11   

cron2 22:11:58
mattock: you're taking notes what we need to fix for 2.2.2?     

mattock 22:12:19
not really, but we should put that stuff to trac
and we got 6 trac tickets already for 2.2.2 22:13:29
anything else besides those? 22:13:35
 
cron2 22:13:48
which ones are the 2.2.2 ones? lost track       

mattock 22:13:58
just a sec
https://community.openvpn.net/openvpn/report/3 22:14:03
scroll back to the bottom 22:14:08
and there's more on the next page 22:14:14
 
cron2 22:14:44
#126 should be fixed now, as soon as we have a signed tap driver        

mattock 22:14:52
I've been reviewing the bug reports and there are quite a few that could 
probably closed        

cron2 22:15:31
#147 is something I wanted to look into 

dazo 22:15:56
#126 I believe will be closed now, with the "small packet fix" from cron2 ... 
#147 and #171 are "outstanding" ... #127, I'd say we either fix quickly before 
release, or postpone it again      

cron2 22:16:22
#154 is hard to fix without configs     

dazo 22:16:41
yeah, I wonder if that's a firewall issue, not openvpn issue    

mattock 22:16:42
dazo: quick fix sounds good     

cron2 22:16:55
ah, indeed      

dazo 22:17:05
#127 isn't that important either        

cron2 22:17:11
"not having client-to-client in the config" just translates to "all packets 
traverse the tun adaptor"
so if there's no firewall on the server host, there's nothing to prevent 
client-to-client config - so this might be a misunderstanding by the user 
22:17:40
 
dazo 22:17:44
exactly 

mattock 22:18:22
close the bug report?   

cron2 22:18:56
I've added a comment explaining this, and since he never sent configs, yes, 
close
well 22:19:06
trac is slow today in accepting comments 22:19:12
 
mattock 22:19:18
I noticed the same
it seems to accept the comments and changes, but hangs forever 22:19:30
if you stop and reload, you'll see the changes 22:19:39
not sure if I should be worried... 22:19:44
seems like google bot is crawling the site 22:21:06
the damn slow Trac git repository, to be exact 22:21:28
so no worries 22:21:31
should we discuss 2.3 briefly? 22:22:47
 
dazo 22:23:49
please, lets do that    

andj 22:24:28
I'm interested  

mattock 22:24:35
so, we have the ticket list here: https://community.openvpn.net/openvpn/report/3
bottom of the page 22:24:39
 
cron2 22:24:45
I'm listening, but I feel a bit exhausted       

mattock 22:24:46
is that up-to-date?     

dazo 22:25:23
pretty much
I did go through a little while ago, where I cleaned up a bit 22:25:41
 
mattock 22:27:30
what's our 2.3 schedule?        

dazo 22:27:41
screwed up?     

mattock 22:27:42
the original schedule went out of the door a while back 
yeah 22:27:44
 
cron2 22:28:33
I want to do more platform cleanups before 2.3  

mattock 22:28:34
I would postpone any major build system changes to post-2.3 period      

cron2 22:28:42
mattock: +1     

mattock 22:28:44
there's "enough stuff" there already    

cron2 22:29:07
let's aim for early feb 2012 for a first 2.3 beta       

dazo 22:29:22
yeah, that sounds reasonable    

mattock 22:29:57
#78 sounds like a more generic problem with <connection> profiles
I recall Jason Haar complaining (rightly) about those in another context 
22:30:11
I also recall that was discussed a long while ago 22:30:21
 
dazo 22:31:27
yeah'   

mattock 22:31:46
can we/should we fix this in 2.3?
we can always make new releases if we feel like it 22:32:01
 
dazo 22:32:47
Yeah,I'm reluctant to say "yes, lets fix it!" ... as we have quite some things 
which needs to be looked at our table
but I'm also that pragmatic that if people don't whine about stuff that often 
... is it that needed to fix such stuff urgently? 22:33:16
#78 has been uncommented for 10 months 22:33:42
(we probably need some bug vote system) 22:33:57
 
mattock 22:34:03
that'd help
then we'd need somebody to fix the upvoted bugs 22:34:21
well, better to fix upvoted bugs than bugs nobody really cares about 22:34:41
 
dazo 22:35:03
exactly 

mattock 22:35:27
#110 would be nice to have, but requires some Windows service knowledge 

dazo 22:35:37
I don't mind letting #78 slip 2.3 ... if it gets noisy, let's pull it into 2.4 
or 2.3.x if urgent       

mattock 22:35:41
and nobody has touched the OpenVPN service code in many, many years
#78 is on the border of "bug fix" and "new feature" 22:36:06
 
dazo 22:36:10
agreed ... #110 would be nice to fix ... is that something d12fk could look at? 

mattock 22:36:11
not sure which one it is        

dazo 22:36:30
considering the few complaints about it .... -> feature 

mattock 22:38:21
added milestone 2.4
moving #78 there 22:38:26
 
dazo 22:38:43
lets see what happens then      

mattock 22:40:36
let's see what else...
#153 is easy fix 22:41:12
I'll do it 22:41:16
#151 and #152 should be doable if 2.3 comes out around Christmas 22:41:47
dazo: you fixed #34? 22:42:20
dazo looks 22:42        

cron2 22:43:15
that's actually a different one, and I'm not sure I understand this
this is more an installer issue than a openvpn.exe runtime thing 22:43:40
 
dazo 22:43:52
it's requested that OpenVPN installation does not update the environment table 
with paths for openvpn/openssl binaries
which might make easy-rsa on windows explode when it doesn't find openssl.exe 
22:44:15
 
mattock 22:44:30
ah, read it too quickly
yes, setpath.nsi does that magic 22:45:20
I can probably fix this easily 22:45:29
 
dazo 22:45:59
if we can avoid updating environment variables, that'd be the very best ... if 
not easy, let's close it
it's not a high profile bug, I believe 22:46:09
 
mattock 22:48:22
well, no-one else has commented on the bug report, so it probably affects a 
limited subset of users
#73 should be easy to fix, simply print out a warning if ccd dir can't be 
opened 22:49:54
cron2 calls it a day for today. wife is starting to get annoyed 22:50   

mattock 22:50:18
what about #106?
cron2: that's always bad 22:50:49
see you later 22:50:55
 
dazo 22:51:20
#73 should be fixed, I believe ... not sure though      

mattock 22:54:48
did not find anything from git logs     

dazo 22:55:37
I know I wrote some patches which did a lot of file path checks
hmmm ... maybe it slipped my own patch queue ... 22:55:50
 
mattock 22:56:19
ok, let's not close it yet, then
then we got two more: https://community.openvpn.net/openvpn/ticket/61 22:56:34
https://community.openvpn.net/openvpn/ticket/106 22:56:41
#61 is being handled by cron2 22:57:04
 
dazo 22:58:38
that might be another pebkac, unless cron2 have discovered something potential 
... could be it has been solved .... no response in 14 months, might indicate 
so 

mattock 22:59:36
in win32.c: fp = fopen ("c:\\windows\\system32\\route.exe", "rb");
also a couple more in the same file 22:59:53
 
dazo 23:00:46
oh crap
well, I only see that one when greping for "c:\" 23:01:46
 
mattock 23:02:43
grep -irn "c...windows" *
gives three matches in win32.c 23:02:57
 
dazo 23:03:39
the one on 108 and 877, I don't dare to touch
but 89, must be fixed 23:03:47
108 and 877 just sets some environment variables, so they're "safer" 23:04:10
 
mattock 23:05:14
mkay, so we're more or less done with the 2.2.1 and 2.3 TODO review
shall we call it a day, I'm getting tired 23:05:20
 
dazo 23:05:31
yeah, lets do that      

mattock 23:05:40
nice!
I hope we get signed TAP-drivers soon, so that 2.2.2 can be released 23:06:12
and I think 2.3 in early January is fully doable 23:06:42
what do you think? 23:06:44
 
krzie 23:06:55
windows tap device got changed again?   

mattock 23:07:34
krzie: a bugfix 

dazo 23:08:02
krzie: small packets on tap driver doesn't go through   

krzie 23:08:44
ahh good to know        

dazo 23:09:14
yeah, cron2 got it fixed, and have done some quick tests ... all looking good   

andj 23:09:52
2.3 in early januari: that would be a beta?     

mattock 23:10:05
signed drivers would allow us to publish a really usable "beta" of 2.2.2        

dazo 23:10:17
I think he meant early february ... but yeah, I'd say beta
(or alpha, if we're not confident enough) 23:10:27
 
andj 23:10:46
cool    

mattock 23:10:59
I meant "first 2.3 alpha in early January"      

dazo 23:11:04
mattock: do we really need to go through betas for minor release?
(2.2.2) 23:11:17
 
mattock 23:11:18
not really
but a preview 23:11:21
 
krzie 23:11:25
whoa 2.3 that soon? 2.3 gets full ipv6 right?   

dazo 23:11:28
yeah    

krzie 23:11:37
you guys are badass
^5 23:11:40
 
dazo 23:11:53
that "soon", our initial idea would have been to have the first alphas already 
released ... but we screwed those plans  

mattock 23:11:55
I don't trust my Windows test boxes anymore, especially when it comes to 
TAP-drivers 
yeah, I think the original release date was last September 23:12:22
 
krzie 23:12:26
its still impressive ;] 

mattock 23:12:36
krzie: if it holds together     

dazo 23:12:52
compared to the 2.1 release .... anything is impressive when it comes to dates 
...      

krzie 23:13:04
lol 2true       

mattock 23:13:35
ok, see you later, got to get some rest 

dazo 23:13:45
2.2 came ~1 year after 2.1, which was good ... and we're keeping that pace, 
which I think is more realistic for our working methods     

mattock 23:13:45
I'll try to summarize this tomorrow     

dazo 23:13:54
mattock: thx a lot!     

mattock 23:14:04
np, good that you made it today
I was afraid the meeting had become a stub 23:14:24
 
dazo 23:14:29

I'm sorry about that ... but life is just too hectic these days, and now the 
body is soon going to protest, so I need to calm down too 23:14:54
 
mattock 23:14:56
one more thing... we should probably quickly review the bug reports with no 
milestone   
dazo even feels he is more grumpier than usually 23:15  

mattock 23:15:20
take your time, we're not really in any hurry   

dazo 23:15:44
mattock: yeah, that's a good point ... I've went through most of of them ... 
and decided to let most of them stay ... some I picked for 2.2.2 and for 2.3, 
but that's about it  

mattock 23:15:58
I look at a few which seemed obsolete   

dazo 23:16:11
yeah, some might be     

mattock 23:16:14
I'll review the rest and we can discuss them in another meeting or unofficially
e.g. one bug with linux 2.4 kernels 23:16:27
dazo goes to try to look for the missing file access check patch before closing 
for today 23:16 

mattock 23:16:32
not confirmed for 2.6
anyways, until next time... take care guys! 23:16:50
 
dazo 23:17:01
2.4 kernels .... uhm ... wontfix ... RHEL3 was the last RHEL release with 2.4 
kernel ... and that's EOL 
dazo just found something nasty in 'master' ... 23:31   

dazo 23:31:29
./openvpn --dev tun ..... this behaves bad      

andj 23:31:55
that's weird, I use that in some of my tests    
dazo need to say it again ... he loves 'git bisect' .... 5 more steps 23:32     

dazo 23:32:53
I'm only starting openvpn with --dev tun ... no more arguments
it's not a valid config ... but it's something I do to test things some times 
23:33:04
 
andj 23:33:11
ah, ok  

dazo 23:33:27
the faulty one spews out:
Thu Nov 24 22:33:15 2011 read from TUN/TAP : File descriptor in bad state 
(code=77) 23:33:28
constantly 23:33:30
the good one just exits 23:33:35
 
andj 23:33:45
heading off as well...  

dazo 23:34:23
enjoy the evening       

andj 23:34:46
thanks  It's 10:30PM here though        

dazo 23:35:00
here too 
hmmm 23:35:53

Reply via email to