[PlanetCCRMA] SOLVED - Seq24 Crashes, EnergyXT2 Not Working in Fedora Core 8

[Fernando Lopez-Lezcano]

On Wed, 2008-04-30 at 02:38 -0700, Nexxus Six wrote:
> To: Fernando (Planet CCRMA) and Jorgen (EnergyXT2):
>
> Tonight I have finally have both Seq24 and EnergyXT2 working fully
> functional in Fedora Core 8 running Planet CCRMA's realtime kernel.

Glad to hear it is working (and sorry for the delay in asnwering)

> In the interest of documenting my work, and what I have done, I am
> writing both lists tonight with the same details, so I don't have to
> repeat a lot of information. 

Seq24 is part of Fedora now, so it would be nice to file a bug report on
the redhat bugzilla tracker. 

Your original report points to a bug in the seq24 program (libc is
detecting an internal inconsistency in the way memory is handled and the
program is stopped). It is possible to turn off those boundary checks in
Fedora with an environment variable which I can't find right now...

That would probably had enabled the program to run, but of course things
were going wrong inside it so it may eventually crash. 

Why was this happening I don't know, most probably one of the underlying
libraries has seen an incompatible upgrade since seq24 was last
compiled. Or seq24 is triggering a bug in the new version of the
library. 

> So, from memory, here's what I've done with my system in the past 24
> hours or so...
> 
> 
> This may sound a bit irrelevant, but one of the applications I had
> recently uninstalled from my system was the game Vega Strike. I had
> the feeling that the reason that neither Seq24 nor EnergyXT2 was
> unable to work (Seq24 crashing when trying to edit a cell, and
> EnergyXT2 freezing when prompted to play) was because of certain
> development libraries missing from my system. I was unable to install
> any of the -devel libraries until after I uninstalled Vega Strike.

Do you remmeber what the error message was?

> Now, I'm not blaming Vega Strike as being the culprit, it just seems
> as though I was unable to add the packages until I deleted Vega Strike
> from my system.
> 
> 
> After the deletion of Vega Strike, I was finally able to add the
> following packages to my system:
> 
> gtk+-devel – 1:1.2.10-59.fc8.i386
> gtk+extra-devel – 2.1.1-7.fc8.i386
> gtk2-devel – 2.12.5-1.fc8.i386
> gtkmm24-devel – 2.12.3-1.fc8.i386
>
> I then was able to successfully compile Seq24 from source code, and
> now have a fully functional version of Seq24 working on my system.
> Prior to this, the ./configure in Seq24 was complaining that my
> Gtkmm-2.4 was not the proper version. I knew this was caused by not
> having the proper -devel files installed.
> 
> Prior to this I had also installed:
> 
> glibmm24 – 2.14.2-1.fc8.i386
> glibmm24-devel – 2.14.2-1.fc8.i386
>
> I then compiled jack.cpp for EnergyXT2's JACK version of libamm.so,
> and now I have a working version of EnergyXT2 running in JACK. I had
> to install the usual GCC, GCC-C++, GCC-Objc, GCC-Objc++, glib,
> glib-devel, glib2, glib2-devel, glibc, glibc-common, glibc-devel,
> glibc-headers libraries to correctly compile the (now working)
> libaam.so file.
> 
> 
> Repositories Used:
> 
> InstallMedia – Fedora Core 8
> fedora – Fedora i386
> livna – Livna for Fedora Core 8 – i386 – Base
> planetccrma – Planet CCRMA 8 -386
> planetcore – Planet CCRMA Core 8 – i386
> updates – Fedora 8 – i386 – Updates
> 
> 
> If either one of you, Fernando or Jorgen, (or anybody on either of the
> lists) needs any additional information, I will try to provide as much
> as I can.
> 
> 
> Question: Is it possible for you as developers of software to somehow
> include all of the extra libraries and whatnot that I had to install
> in order to get all of this working “out of the box” per se? 

No, not really. Or yes, it is already done - at two different levels :-)

When you install a package like seq24 yum installs all the libraries
that the package needs to run because the package tells yum what it
needs. But the package dependencies for seq24 do not include what you
would need to _compile_ seq24 from scratch. It would not make sense to
do that at all. 

Now, if you try to rebuild the _source_ rpm package (.src.rpm) to create
binary rpms from scratch, then rpmbuild will definitely complain if
there are any build dependencies missing and will tell you what they are
- I mean the name of the actual packages that are missing (this is done
in the spec file for the package in question through the BuildRequires:
directive). 

At a different level the configuration process for seq24 should also
tell you what is missing, but it will not be so explicit as seq24 itself
(the source) does not know anything about how things are packaged in
Fedora. So it will tell you that, say, a certain .h file is missing and
from there you will have to figure out what Fedora development package
provides it (yum can also help with that). 

> It just seems like a lot of extra work for something that should run
> otherwise... 

It does if you try to rebuild the .src.rpm package. 

> I'm not complaining here, but it would be nice if the install program
> could be a little more verbose in its requirements. Seq24, for
> example, installed with no problems from Fedora's RPM Installer, but
> made no indications that I was going to need a lot of -devel and other
> low-level support libraries and applications. 

You would know that from the .src.rpm build requirements. 

> I guess this is something that Windows Users take for granted...

I imagine most Windows users don't build packages at all...

> One thing I noticed was the size difference in Seq24 after I
> recompiled it... going from 330k, to nearly 4.5Mb in size. What
> accounts for the difference?

Built in debugging information that is split into a different package in
the case of Fedora (that would be the seq24-debuginfo package - you
normally don't need all that stuff to be able to run the program). 

> At any rate, everything is now working, and I'm going to finally get
> back to doing what I love the most: Making some music! :-)
> 
> 
> Questions? Comments? Suggestions?

Please file a bug report in bugzilla.redhat.com... (against the seq24
package). Thanks for the detailed report!

-- Fernando