Linux users of 1.2.5-CVS, feedback, please!

[t00fri]

I am mystified!

Feedback from /Linux users/ that already use the
1.2.5-CVS code would be very helpful:

The problem I noticed in my SuSE 7.3 Linux installation (Logitech
Pilot Optical Wheel mouse) at home, is that the "click object & display" in the
majority of cases (Tycho stars from Rigel's stardb-1.2) does not work
anymore with a /freshly/ installed

1.2.5-CVS Celestia version.

In more detail:

1) Suppose I zoom into small FOV's (e.g. 1deg) by means of "SHIFT mouse
button L" and click on some faint stars, /nothing/ is displayed in the
upper left corner. It works, however, with planets and a number of
brighter stars.

2) In my office machine, with the /identical/, freshly installed
1.2.5-CVS, but an ATI card (instead of my GeForce 2GTS
at home) and SuSE7.1 (instead of SuSE 7.3), the "click object & display"
works perfectly!

3) At home, a freshly installed 1.2.4 (final) /also/ works perfectly!!
Note it uses the same Celestia data as the 1.2.5-CVS in /usr/share/celestia.
If I just add the Tycho modifications in stardb.cpp and
celestiacore.cpp, the Tycho star numbers are also correctly
displayed.

So something is fishy after the more recent code checkins. However, tracing it
is difficult since, apparently, things are system dependent! Moreover,
most of the new code is the same for Windows and Linux. And things
work fine also in my Windows ME installation.

That's why I need your feedback.

Note, everything still works perfectly as of my last code checkin
before my vacations, on June 22nd.

Chris has promised to check this immediately quite a number of days ago, but
I have not heard from him about the outcome since;-) (The "Nirvana" syndrome)
He was the only one who checked in code after June 22nd, as far as it seems...

Thanks for your help

Bye Fridger

[chris]

Chris has promised to check this immediately quite a number of days ago, but
I have not heard from him about the outcome since;-) (The "Nirvana" syndrome)
He was the only one who checked in code after June 22nd, as far as it seems...

I did check it immediately, but I forgot to post about it . . . I updated from CVS and am having troubles getting star names to display correctly. I think that Celestia is confused about the cutoff magnitude--it thinks most stars are too faint to be selectable.

--Chris

[t00fri]

Chris has promised to check this immediately quite a number of days ago, but
I have not heard from him about the outcome since;-) (The "Nirvana" syndrome)
He was the only one who checked in code after June 22nd, as far as it seems...

I did check it immediately, but I forgot to post about it . . . I updated from CVS and am having troubles getting star names to display correctly. I think that Celestia is confused about the cutoff magnitude--it thinks most stars are too faint to be selectable.

--Chris

Aha!

But then one might immediately ask, why Celestia is not confused about the cutoff magnitude in Version 1.2.4 with /exactly/ the same data?

Bye Fridger

PS: Chris, I just got another email back that I sent to you Sat, Sept 07, about dds bump file problems.

Once more, the mailer tried for 5 days without success to reach claurel@ns.shatters.net.

[t00fri]

I did check it immediately, but I forgot to post about it . . . I updated from CVS and am having troubles getting star names to display correctly. I think that Celestia is confused about the cutoff magnitude--it thinks most stars are too faint to be selectable.

--Chris

Your remark, indeed, put me on the right track.

I have located and fixed the "click object & display" problem for Linux-1.2.5-CVS. It arose from a missing sync in gtkmain.cpp in case of AutoMag, where 'faintestVisible' may change within the simulation due to FOV zooming.

The needed changes are not entirely trivial, though. After some further testing, I will check it in along with some improvements of the AutoMag scheme in view of the new 780K star data...

Bye Fridger