SVN compiles but segfaults at splashscreen

[HScoop]

I tried compiling the latest SVN using Ubuntu 6.10 according to the instructions in the Wiki. It compiled without much of a problem but when I try to run it, it crashes at the splashscreen with a segfault.

I'm having the same problem using the latest SVN and the last download version (10722), although the stable version did run without any problems, so I'm assuming I messed up compilation somehow - unfortunately I'm pretty clueless in that area...

I've attached the output of GDB and config.log, don't know if there's anything else you could use.

[ace123]

So it's pretty clear that the crash is in TextPlane::Draw, hud.cpp:232

Code: Select all

  if (!use_bit) {
    scalex=(_Universe->numPlayers()>3?_Universe->numPlayers()/2:_Universe->numPlayers())*myFontMetrics.i/std_wid;
    scaley=myFontMetrics.j/(119.05+33.33);
  }

The only thing that would cause a crash in an accessor function like numPlayers() would be if _Universe is NULL.

Now, this happens to be the only place where it references _Universe, and it only happens if not using bitmap fonts. A NULL _Universe is an easy condition to check for, so I'll commit a fix with that.

Well I did just set high_quality_fonts to "false" in the config file, and immediately I got the crash... at some point, the loading order must have changed... not that this is a bad thing, since the sooner it can get a loading graphic up, the better.

Even more worrisome is that "svn blame" pointed out that I was the one who committed this very line of code almost 4 years ago...

Any way, I am committing the fix for this. Thanks for debugging this one.

[HScoop]

Thanks for the quick response and fix - it works neatly now

[klauss]

I made the splash show sooner, so sorry, my bad.
I wasn't expecting an access to Universe for rendering text...

[ace123]

I'm glad you did that... that long delay before the splash screen has been the source of many complaints, bug reports and uninstalls.

The main reason it hadn't been done was because of all the calls to _Universe... I guess it turns out that there weren't as many as one would have thought.