Building configparser binary fails.

[Thijs van Veen]

To whom it may concern,

I'm using OLSR v. 0.6.1 on Ubuntu 12.04 (3.2.0-45 kernel).

When trying to build the binary version of cfgparser, make exits with
errors.
The library version compiles perfectly fine, as does OLSR itself.

I tried the following two invocations (seeing as the Makefiles use
simple ifdefs, the value shouldn't matter):
***@user-laptop:~/olsr/olsrd$ make cfgparser MAKEBIN=1
***@user-laptop:~/olsr/olsrd/src/cfgparser$ make MAKEBIN=1

I've saved the output from the second invocation to:
http://pastebin.com/Mu6d5gdH

The first output is similar, but with slightly different paths (as is to
be expected).
--
Thijs van Veen
***@operamail.com

[Ferry Huberts]

Post by Thijs van Veen
To whom it may concern,
I'm using OLSR v. 0.6.1 on Ubuntu 12.04 (3.2.0-45 kernel).
When trying to build the binary version of cfgparser, make exits with
errors.
The library version compiles perfectly fine, as does OLSR itself.
I tried the following two invocations (seeing as the Makefiles use
http://pastebin.com/Mu6d5gdH
The first output is similar, but with slightly different paths (as is to
be expected).

I think you're the first to notice (we didn't...) this since a very long
time.
From which I conclude that nobody uses it this way.

So if it's important enough to you then I'm inviting you to supply
patches to fix this build :-)

[Henning Rogge]

Post by Thijs van Veen
To whom it may concern,
I'm using OLSR v. 0.6.1 on Ubuntu 12.04 (3.2.0-45 kernel).
When trying to build the binary version of cfgparser, make exits with
errors.
The library version compiles perfectly fine, as does OLSR itself.
I tried the following two invocations (seeing as the Makefiles use
http://pastebin.com/Mu6d5gdH
The first output is similar, but with slightly different paths (as is to
be expected).

Can you try to attached patch? (only compile-tested)

Henning Rogge

[Ferry Huberts]

Post by Henning Rogge

Post by Thijs van Veen
To whom it may concern,
I'm using OLSR v. 0.6.1 on Ubuntu 12.04 (3.2.0-45 kernel).
When trying to build the binary version of cfgparser, make exits with
errors.
The library version compiles perfectly fine, as does OLSR itself.
I tried the following two invocations (seeing as the Makefiles use
http://pastebin.com/Mu6d5gdH
The first output is similar, but with slightly different paths (as is to
be expected).

Can you try to attached patch? (only compile-tested)

it fixes that build
shall I commit it with you as author?

[Henning Rogge]

Post by Ferry Huberts

Post by Henning Rogge
Can you try to attached patch? (only compile-tested)

it fixes that build
shall I commit it with you as author?

I would like to see first if the patch solves the problem. Just keeping
the compiler quiet is nice, but lets give it a few days to see if its
enough.

Henning Rogge

[Thijs van Veen]

Thanks for the replies.
I just stumbled upon the bug and don't really need the standalone parser
per se, but I thought it would be nice to at least report the bug.

It compiles indeed after patching, but I do get a segmentation fault
when trying to run it though.
I've posted gdb output + configuration to: http://pastebin.com/t04UxRjK.

Note: olsrd runs perfectly (as far as I know) with that configuration
file and I've made sure I've got permissions to read and even modify the
fle.

Like I said, I don't really need the binary, so there's no hurry here,
but I assumed you guys want the program to be bug-free as well.
--
Thijs van Veen

Post by Henning Rogge

Post by Ferry Huberts

Post by Henning Rogge
Can you try to attached patch? (only compile-tested)

it fixes that build
shall I commit it with you as author?

I would like to see first if the patch solves the problem. Just keeping
the compiler quiet is nice, but lets give it a few days to see if its
enough.
Henning Rogge
--
Diplom-Informatiker Henning Rogge , Fraunhofer-Institut für
Kommunikation, Informationsverarbeitung und Ergonomie FKIE
Kommunikationssysteme (KOM)
Fraunhofer Straße 20, 53343 Wachtberg, Germany
Telefon [phone], Fax [phone]
--
Olsr-users mailing list
https://lists.olsr.org/mailman/listinfo/olsr-users
+ smime.p7s
8k (application/pkcs7-signature)

[Henning Rogge]

Post by Thijs van Veen
Thanks for the replies.
I just stumbled upon the bug and don't really need the standalone parser
per se, but I thought it would be nice to at least report the bug.
It compiles indeed after patching, but I do get a segmentation fault
when trying to run it though.
I've posted gdb output + configuration to: http://pastebin.com/t04UxRjK.

I pushed a fix that solve (for me) both the compilation and the segfault
issue. Maybe you want to test it again.

Henning Rogge

[Ferry Huberts]

Thijs, Henning,

I added 2 makefile targets for cfgparser and our Jenkins now also builds
the standalone cfgparser in all our builds (linux64, linux32, android,
win64, win32, all in debug and non-debug mode, so 10 builds).

Now we'll see right away when the cfgparser build falls over, thanks Thijs!

Post by Henning Rogge

Post by Thijs van Veen
Thanks for the replies.
I just stumbled upon the bug and don't really need the standalone parser
per se, but I thought it would be nice to at least report the bug.
It compiles indeed after patching, but I do get a segmentation fault
when trying to run it though.
I've posted gdb output + configuration to: http://pastebin.com/t04UxRjK.

I pushed a fix that solve (for me) both the compilation and the segfault
issue. Maybe you want to test it again.
Henning Rogge

[Thijs van Veen]

Ferry, Henning,

Nice to see things get picked up so fast, however I did find two things:

1) Building still seems to have the same behaviour, where the cfgparser
(shared lib and binary) is not automatically built.

I made sure I'm running the latest version (checked the commitdiff and
local file). Defining the OS does not seem to influence it.
Also, I'm not sure whether it's by design, but the cfgparser binary and
shared lib don't get cleaned up when running "make clean" from the olsrd
root dir, while the object files do get cleaned.

2) I found a new segmentation fault when running the cfgparser binary.

The segmentation fault occurs when running cfgparser with the -print
option.
The output can be found here: http://pastebin.com/9XVpGb8P
As can be seen in the paste, running the parser without -print seems to
work fine.
--
Thijs van Veen
***@operamail.com

P.S. These days I'm hanging around in #olsr on freenode IRC in case you
prefer direct contact for solving/testing these kind of issues.

Post by Ferry Huberts
Thijs, Henning,
I added 2 makefile targets for cfgparser and our Jenkins now also builds
the standalone cfgparser in all our builds (linux64, linux32, android,
win64, win32, all in debug and non-debug mode, so 10 builds).
Now we'll see right away when the cfgparser build falls over, thanks Thijs!

Post by Henning Rogge

Post by Thijs van Veen
Thanks for the replies.
I just stumbled upon the bug and don't really need the standalone parser
per se, but I thought it would be nice to at least report the bug.
It compiles indeed after patching, but I do get a segmentation fault
when trying to run it though.
I've posted gdb output + configuration to: http://pastebin.com/t04UxRjK.

I pushed a fix that solve (for me) both the compilation and the segfault
issue. Maybe you want to test it again.
Henning Rogge

--
Ferry Huberts

[Thijs van Veen]

Quick update:

I found another build fail when trying to build the binary after the
shared lib has been made. The other way around does work.
The output: http://pastebin.com/8CVLMxkE
--
Thijs van Veen

Post by Thijs van Veen
Ferry, Henning,
1) Building still seems to have the same behaviour, where the cfgparser
(shared lib and binary) is not automatically built.
I made sure I'm running the latest version (checked the commitdiff and
local file). Defining the OS does not seem to influence it.
Also, I'm not sure whether it's by design, but the cfgparser binary and
shared lib don't get cleaned up when running "make clean" from the olsrd
root dir, while the object files do get cleaned.
2) I found a new segmentation fault when running the cfgparser binary.
The segmentation fault occurs when running cfgparser with the -print
option.
The output can be found here: http://pastebin.com/9XVpGb8P
As can be seen in the paste, running the parser without -print seems to
work fine.
--
Thijs van Veen
P.S. These days I'm hanging around in #olsr on freenode IRC in case you
prefer direct contact for solving/testing these kind of issues.

Post by Ferry Huberts
Thijs, Henning,
I added 2 makefile targets for cfgparser and our Jenkins now also builds
the standalone cfgparser in all our builds (linux64, linux32, android,
win64, win32, all in debug and non-debug mode, so 10 builds).
Now we'll see right away when the cfgparser build falls over, thanks Thijs!

Post by Henning Rogge

Post by Thijs van Veen
Thanks for the replies.
I just stumbled upon the bug and don't really need the standalone parser
per se, but I thought it would be nice to at least report the bug.
It compiles indeed after patching, but I do get a segmentation fault
when trying to run it though.
I've posted gdb output + configuration to: http://pastebin.com/t04UxRjK.

I pushed a fix that solve (for me) both the compilation and the segfault
issue. Maybe you want to test it again.
Henning Rogge

--
Ferry Huberts

--
http://www.fastmail.fm - A no graphics, no pop-ups email service
--
Olsr-users mailing list
https://lists.olsr.org/mailman/listinfo/olsr-users