Re: Maintaining Jami #3

2020-02-07 Thread Jan
On Fri, 07 Feb 2020 23:34:24 +0100
Marius Bakke  wrote:

> Nice find!  The fix[0] seems straightforward, would it work to
> download the patch[1] and apply it for pjproject 2.9?
> 
> [0] https://trac.pjsip.org/repos/changeset/6087
> [1] https://trac.pjsip.org/repos/changeset/6087?format=diff=6087

I applied the patch and it resolved the bug, when jami crashes during
video call! Got another crash, when I pressed "disconnect" button, but
guess that's just standard Jami experience, anyway one step forward!
I think more bugs will be resolved after the 2.10 release of pjproject
(they fix lots of stuff there).
I'll clean up my repo tomorrow and post changes.


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-02-07 Thread Jan
Dnia 2020-02-07, o godz. 23:34:24
Marius Bakke  napisaƂ(a):

> Nice find!  The fix[0] seems straightforward, would it work to
> download the patch[1] and apply it for pjproject 2.9?
> 
> [0] https://trac.pjsip.org/repos/changeset/6087
> [1] https://trac.pjsip.org/repos/changeset/6087?format=diff=6087

I've been thinking about this. What's the procedure of applying patches
in Guix? Should I use my procedure? How to copy the patch into the
package source tree? Or should I use substitute*?


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-02-07 Thread Marius Bakke
Jan  writes:

> Hello,
> I think I found the source of the problems - there's a bug in pjproject
> 2.9
> https://trac.pjsip.org/repos/ticket/2240
> and 2.10 wasn't yet released, so I can't package it.
> The error message is the same as I got during testing calls.
> I guess the bug have been unnoticed by Jami developers, because:
> 1. their builds are not reproducible
> 2. software is tested and used in different network environments -
> there's a chance specific settings trigger the bug

Nice find!  The fix[0] seems straightforward, would it work to download
the patch[1] and apply it for pjproject 2.9?

[0] https://trac.pjsip.org/repos/changeset/6087
[1] https://trac.pjsip.org/repos/changeset/6087?format=diff=6087


signature.asc
Description: PGP signature


Re: Maintaining Jami #3

2020-02-07 Thread Jan
On Fri, 07 Feb 2020 14:53:23 +0100
Pierre Neidhardt  wrote:

> This is great news, thanks for keeping up with the hard work!
> 
> At this point, I think it's safe to say that Jami is still a
> work-in-progress :p
Yeah, it's always wip in my opinion :D

> Jan, what do you think of making a channel of your work with Jami, so
> that other users can easily benefit from it and test / report?
> This could be useful until things settle down.
> 
You mean a guix channel? I must read more about channels and
clean up my repository on notabug dot org. I also must finish one git
chapter about branching, to prevent making a huge mess when rebasing
and merging.



Jan Wielkiewicz



Re: Maintaining Jami #3

2020-02-06 Thread Jan
Hello,
I think I found the source of the problems - there's a bug in pjproject
2.9
https://trac.pjsip.org/repos/ticket/2240
and 2.10 wasn't yet released, so I can't package it.
The error message is the same as I got during testing calls.
I guess the bug have been unnoticed by Jami developers, because:
1. their builds are not reproducible
2. software is tested and used in different network environments -
there's a chance specific settings trigger the bug

Until the release of pjproject 2.10 I can do nothing, but I actually
cleaned up my private guix repository, so DO NOT pull my changes,
because they're a total mess. I guess I'll close the current issue and
open a new one once I finish cleaning my code. I'll also finish
learning git, as I have some free time now.



Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-27 Thread Jan
Hi Pierre,
I made some tests Guix System <-> Devuan on my current Jami package
20200124 version or something similar to this and I managed to
reproduce bugs you pointed out.
Didn't get answer from Jami developers yet, but it seems something is
wrong with our pjproject package:
That's what the daemon prints while crashing:
dring: ../src/pjsip-ua/sip_inv.c:246: pjsip_inv_dec_ref: Assertion `inv
&& inv->ref_cnt' failed.

Now I am going to investigate what's wrong with pjproject, mostly
finish our currently not working pjproject package (it won't compile,
only pjproject-jami works) and then I'll update it to the latest
version, if it got outdated much since the last time I checked.

I also wait for merging core-updates, because another bug - the
screen sharing one may be caused by something related to glibc, which
got updated on core-updates. Or webkit-gtk or Qt or whatever - anything
can be a problem, if the package has about 1GB of dependencies...


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-21 Thread Pierre Neidhardt
Jan  writes:

> I didn't really answer your questions earlier, sorry.
>
> On Sun, 19 Jan 2020 10:20:51 +0100
> Pierre Neidhardt  wrote:
>
>> Hi Jan,
>> 
>> I tested the Jami package we have in upstream Guix:
>> 
>> - It fails to start on my desktop, it only works on my laptop.
> What do you mean by desktop, is it even a difference for the OS and the
> application?

I have a desktop computer and a laptop.  Both running a mostly identical
Guix system.
The hardware is different; in particular, the desktop does not have a
mic or a camera, if that matters.

>> - The client kept disconnecting.  I could send a few text messages
>> here and there, but most of them didn't go through.
> For me it works, I test it in my local network though. The problem
> begins when you try communicating with someone behind a strong NAT,
> firewall or a badly configured network. Based on my previous experience
> I think that's normal - it have never worked well enough, because ISPs
> don't want you to have control over your networking - client-server
> applications work well, but p2p are broken. Often router firmware is
> buggy, outdated and malicious as nonfree software is.

I've tried between my laptop and an Android.  Both were on the local network.

>> - I can receive calls, but when I click on the "pick up" button, Jami
>>   crashes.
> Maybe its something hardware specific or depends on the client you call
> with - I have no problem answering calls made from Android.
>> In general, Jami crashed _a lot_.  I wonder if it's upstream or if we
>> have something wrong with our package.
>> 
>> Did you have a similar experience?
>> Does your work fix any of those issues?
>> 
>
> I tested 20200115 version now, and it pretty much works the same.
> Anyway I fixed my private branch and can continue working, but see no
> point in waiting.
>
> What DE are you using? I could try it on my machine.

I run EXWM.

> If you run it on top of Wayland, then you should know Jami is
> currently broken on Wayland, they work on fixing it. Aaaand even if
> client-gnome will continue being broken like this, they're working on
> porting client-windows(Qt) to GNU/Linux and there's a possibility it
> won't be broken as well.

That'd be interesting.

-- 
Pierre Neidhardt
https://ambrevar.xyz/


signature.asc
Description: PGP signature


Re: Maintaining Jami #3

2020-01-20 Thread Jan
I didn't really answer your questions earlier, sorry.

On Sun, 19 Jan 2020 10:20:51 +0100
Pierre Neidhardt  wrote:

> Hi Jan,
> 
> I tested the Jami package we have in upstream Guix:
> 
> - It fails to start on my desktop, it only works on my laptop.
What do you mean by desktop, is it even a difference for the OS and the
application?

> - The client kept disconnecting.  I could send a few text messages
> here and there, but most of them didn't go through.
For me it works, I test it in my local network though. The problem
begins when you try communicating with someone behind a strong NAT,
firewall or a badly configured network. Based on my previous experience
I think that's normal - it have never worked well enough, because ISPs
don't want you to have control over your networking - client-server
applications work well, but p2p are broken. Often router firmware is
buggy, outdated and malicious as nonfree software is.

> - I could not give an audio or video call
Works on my machine, except for the bug caused by screen sharing. 

> - I can receive text messages perfectly.
This one works.

> - I can receive calls, but when I click on the "pick up" button, Jami
>   crashes.
Maybe its something hardware specific or depends on the client you call
with - I have no problem answering calls made from Android.
> In general, Jami crashed _a lot_.  I wonder if it's upstream or if we
> have something wrong with our package.
> 
> Did you have a similar experience?
> Does your work fix any of those issues?
> 

I tested 20200115 version now, and it pretty much works the same.
Anyway I fixed my private branch and can continue working, but see no
point in waiting.

What DE are you using? I could try it on my machine. If you run it on
top of Wayland, then you should know Jami is currently broken on
Wayland, they work on fixing it. Aaaand even if client-gnome will
continue being broken like this, they're working on porting
client-windows(Qt) to GNU/Linux and there's a possibility it won't be
broken as well.


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-19 Thread Jan
On Sun, 19 Jan 2020 10:20:51 +0100
Pierre Neidhardt  wrote:

> Hi Jan,
> 
> I tested the Jami package we have in upstream Guix:
> 
> - It fails to start on my desktop, it only works on my laptop.
> 
> - The client kept disconnecting.  I could send a few text messages
> here and there, but most of them didn't go through.
> 
> - I could not give an audio or video call.
> 
> - I can receive text messages perfectly.
> 
> - I can receive calls, but when I click on the "pick up" button, Jami
>   crashes.
> 
> In general, Jami crashed _a lot_.  I wonder if it's upstream or if we
> have something wrong with our package.
> 
> Did you have a similar experience?
> Does your work fix any of those issues?
> 

Well, generally I started working on Jami, because it never worked for
me well - crashes, bugs, issues with UPnP and NAT. That was on Devuan,
but on Guix it is like this also. I think bugs are present upstream,
but Guix only makes them reproducible :)
I also don't know which releases are beta and which are not, I once
tried asking, but did not get real answer.
Their release naming system is MMDD and it tells you nothing, I
asked them to change it, but they didn't. 
I just package the latest version available at
https://dl.jami.net/ring-release/tarballs/?C=M;O=D
and sometimes its okay, but once I couldn't build gnome client on
Devuan at all.

Jami should not be considered stable for now, at least my package - they
make too many changes for it to be stable, without any code freeze.
The software was started in 2016 and two years ago it was a deep beta,
they need some time to clean this up.

Current versions also use beta software - OpenDHT 2.0 is currently at
beta stage.

Once my changes are merged, I can try updating everything to the latest
versions and maybe it'll work, if not we can rename jami to jami-beta
and rename it back when it's ready.



Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-18 Thread Jan
I'm really sorry for this spam, but the GSM we have packaged isn't the
GSM I'm talking about, didn't notice this.
Help me identifying the license please, sorry for the confusion.
How should I name the package, if both have the same name?


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-18 Thread Jan
Nevermind, I somehow missed GSM is already packaged, sorry.
Anyway, the second question is still relevant.


Jan Wielkiewicz




Re: Maintaining Jami #3

2020-01-18 Thread Jan
Hi,
Although pjproject-jami works, I decided to fix the broken pjproject
package. In order to do this, I need to package GSM, but I'm not sure
what license is it using, can someone help me idendifying the
following license:

-
Copyright 1992, 1993, 1994 by Jutta Degener and Carsten Bormann,
Technische Universitaet Berlin

Any use of this software is permitted provided that this notice is not
removed and that neither the authors nor the Technische Universitaet Berlin
are deemed to have made any representations as to the suitability of this
software for any purpose nor are held responsible for any defects of
this software.  THERE IS ABSOLUTELY NO WARRANTY FOR THIS SOFTWARE.

As a matter of courtesy, the authors request to be informed about uses
this software has found, about bugs in this software, and about any
improvements that may be of general interest.

Berlin, 28.11.1994
Jutta Degener
Carsten Bormann
-

I guess it's "license:non-copyleft", but I'm not sure.

By the way, have someone checked my commits for Jami? My private repo
starts getting more and more outdated, what makes it harder for me to
continue working, because I'm not a git wizard yet.


Jan Wielkiewicz



Re: Maintaining Jami #3

2020-01-07 Thread Jan
The issue is now available here 
https://issues.guix.gnu.org/issue/39005


Jan Wielkiewicz