Re: [gccsdk] autobuilder - GLib-Critical error

Hi Theo

Starting in an xterm causes firefox to be popped open when the glib err is reported, then the script continues.

the condensed (# comments removed) setvars equates to

echo "1"

AB_HG="https://..."

echo "2"

(where ... is the correct hg path)

THEN:
autobuilder, via fetch-program, scans setvars as a file and correctly does the mercurial clone (its about a GB), after which it invokes setvars as a script

setvars, once invoked, prints "1" to the terminal then pops up the glib error.

in a basic login it stops here. In an X terminal window firefox is started with no helpful info and the script continues by printing "2"

i.e. a basic failure within the build system before it does anything, really, other than invoke a shell script following a mercurial checkout. 

At least I know what is going on.. I suspect the attempt to open firefox is the issue that gives the glib error, though why that is happening is unclear

Cheers

John
 

On 18/05/2015 04:03, Theo Markettos wrote:

On Sun, May 17, 2015 at 09:50:10PM +0100, John Ballance wrote:  

Hi Theo    Thanks for the reply, and for the suggestions    A few points    1: Though there is a ubuntu desktop around (ctrl-alt-F7) I am  working directly from a plain text (?) login at ctrl-alt-F1    2: What I said about the failure is accurate.    the setvars script has the AB_HG line in it, which is detected and  used by autobuilder fetch-program script. Once that has completed  the sources are in place. The autobuilder then invokes setvars. It  runs fine (doing echo 'some text') until it encounters the AB_HG=  line, at which point the GLib error ss reported and nothing goes  further.  

  But the AB_HG line is just setting an environment variable, it doesn't  execute anything?    What does the setvars file look like (what's the package, if it's one in the  tree?)    

I've now tried again in a desktop terminal window. GLib throws all  sorts of funnies out, eventually returning to setvars to continue...  

  I think you'll need to post the log - often what you see is a symptom of an  earlier problem, and without the full log we don't have much to go on.    

I tried the code you suggested for Xvfb, but it then complains it  cannot open display 99 (your example).    I presume Xvfb :99 needs an & after it to return to command line?  

  Yes.    

I'm still none the wiser as to why glib decided it needed to push  something to the display though, and none the wiser what it actually  tries to push.  

  I have no idea.  It's unusual for things to need X during builds, and it's  also unusual for things to invoke gtk/glib apps as part of the build.  The  log should tell us more.    Theo    

--

Stop press! Just announced -

Most Trusted Broadband Provider in the 2014 Moneywise Customer Services Awards

For full details of see https://www.utilitywarehouse.co.uk/reviews?exref=095761

Intrigued? Call me

John Ballance C.Eng MIET - jwb@macpcrepair.co.uk - 07976 295923