$Id: BUGS,v 1.9 2003/04/23 14:37:57 d3august Exp $

Fixed bugs are "V"ed. Unfixed have stars. 
"Z"s are not being actively worked on.

V infinity.nus.sg goes all the way up to 30 hops, and the last four or so
  are "* * *". Dumps.
V Start a new trace while another one is underway -> weird results.
V When the DNS doesn't answer, the program hangs.
V Sometimes I get stuck in the dragmode. 
V Earth isn't centered when the glarea is very non-square.
V Selecting yellow sites won't affect them on the globe.
V The entries in the table in the infowindow are centered. They shouldn't be.
V last sites if not known are unselectable. Ex: www.kei.com
V An error in placement *sometimes*. (It was in the database-reading, 
  uninitialized values *sometimes* used.
* Some kind of rounding error or something in locStrToNum <-> locNumToStr
  somewhere. This makes writing db's very shaky.
V Add the hashes to the comments in the networks files.
V The Help menu is not at the far right of the window anymore.
  (It is now. Was a gtk bug.)
V Weird distance-values on SGI. (Was a float<->double rounding error.)
Z Remove all // comments, or fix makefile.
V On some OpenGL implementations, the sites are visible thorough the earth.
  This happens on (at least) SGI onyx2, mesa. (GL_DEPTH_SIZE,1, Duh!)
V sgi: traceroute prints errors to stdout. -> dump on "host not found."
V picking doesn't work correctly when the window has been resized.
V The "Help"-menu isn't at the far right anymore, about a cm in.
     (fixed by new GTK version)
V only the Yes button on the new trace dialog works, not pressing enter.
V Dumps on linux. (Was a race-condition in the compiler code-layout.)
V Dumps on linux when pressing OK in the "new trace" dialog. (Was _another_
  popen()-inconsistency! (This time: different behavior on Linux and Solaris
  WRT what happens then you pclose(NULL). (solaris returns -1, linux SIGSEGV.)
  (This is apparently a glibc bug. (Confirmed, known bug.))
V _Something_ is wrong with malloc/freeing somewhere. Dump under linux 
  connected to three frees in main.c. (Went away. Must have been a buggy 
  libc)
V When it can't find traceroute, I get some gtk warnings, and a window
  that tells me that it can't find the site.
V If the last site is unknown, it ends up on (0,0).
V xtraceroute sdkhgsginsdb -> ** WARNING **: invalid cast from (NULL) pointer to `GtkLabel
* Core dump in Suns libc (gettext) if I try to use it. Go with GNU
  gettext for now.
V Weird garbage in the beginning of some strings sometimes. (Fixed.
  Don't strcat to a new string, strcpy first time.)
V Handling of the "* * *" case on linux is wrong. It seems to hang
  there, spinner running.
V Doesn't plot the last site on linux.
V Clicking in the glarea to select doesn't always work on linux.
  (Resizing the window helps.) (Happens on Solaris as well, since I
  changed to Windowmaker. Related?)
  Was order of calls to resize / realize callbacks. Changed to
  reconfig_event, works...
V Order of the -l<lib> entries on the link line is wrong. (-lgtkgl 
  after -lGL)
V Using snprintf, which isn't available everywhere. (g_snprintf)
V "xtraceroute --stdin < apa" works, "cat apa | xtraceroute --stdin"
  doesnt.
V --stdio? Should be --stdin
V The initial view should be centered on the user's location, not
  looking at (0,0).
* Report of weird colors (blue ground, brown sea) on solaris.
V Async lookup of first site (to rotate the globe correctly) doesn't
  work. (Does now, after a mild hack to get_from_extpipe().)
V file->new doesn't work.
* Sites can end up under a sat jump, like with a normal link. Find out
  on which side it is using the ping time.
* Starting a new trace while another one is in progress doesn't work
  anymore. Fix clear_sites.
V It fails to detect when traceroute says "unknown host"
V Going from chalmers to www.gov.cl gives me a negative spinner
  refcount.
V Fix the perspective. It should work with non-quadratic canvases.
V selection should work even if the canvas has been resized...
* Spinner refcount negative...
V click info on a red/orange site -> acc: n/a, forward, back, it changes.
V Window can't be made smaller than the default starting size.
V Only nag the user once about not using the EW version of "host".
* Configuring with solaris cc screws up the makefile for some reason.
V The resolver script doesn't work for boxes with no domain (a hostname
  with no dots) in the BIND case.