put_comb · brlcad · Zulip Chat Archive

Stream: brlcad

Topic: put_comb

starseeker (Sep 04 2020 at 19:17):

Hey @Sean , is there any chance that there is some uncommitted work from the comb dialog refactor? src/tclscripts/mged/comb.tcl@970 has a put_comb call with 10 args, but it looks like the libged/put_comb/put_comb.c code doesn't have a case for that...

Sean (Sep 04 2020 at 19:18):

let me look

Sean (Sep 04 2020 at 19:20):

I'm only seeing 9 args, or are you counting the command name as an arg also?

starseeker (Sep 04 2020 at 19:21):

put_comb
$comb_control($id,name)
$color
$comb_control($id,shader)
$comb_control($id,inherit)
$comb_control($id,comb)
$comb_control($id,isRegion)
$comb_control($id,id)
$comb_control($id,air)
$comb_control($id,material)
$comb_control($id,los)

Sean (Sep 04 2020 at 19:22):

yeah, I just caught that -- missed color

Sean (Sep 04 2020 at 19:22):

usage on put_comb says that's supposed to be correct

starseeker (Sep 04 2020 at 19:23):

Hmm... Wonder if I messed something up in the refactor - put_comb.c code doesn't seem to cover it...

Sean (Sep 04 2020 at 19:24):

still refreshing my memory on this. I know I swapped one of the args around.

Sean (Sep 04 2020 at 19:28):

yeah, so it was calling put_comb with 10 args before the swap in r76299 too, checking the impl..

Sean (Sep 04 2020 at 19:28):

woah, what happened to the history? how did you move the files into subdirs?

Sean (Sep 04 2020 at 19:29):

oh, there we go.. whew that was weird.

Sean (Sep 04 2020 at 19:30):

I thought svn log on a dir showed history on all contents, but apparently not. had to history the file itself.

starseeker (Sep 04 2020 at 19:30):

/me has actually started using the git conversion for history lookups

starseeker (Sep 04 2020 at 19:31):

/me is liking gitk, on the whole

starseeker (Sep 04 2020 at 19:35):

(I've been periodically updating the conv10 repo on github (not final form, haven't been regenerating the account mappings) so it's current-ish on the history if it's helpful.)

Sean (Sep 04 2020 at 19:39):

yeah, totally hoping to get back to that as soon as these reviews are done. nearly there.

Sean (Sep 04 2020 at 19:39):

okay, refreshed on the change and it's what I recalled -- the last four args moved to before the region flag, but not args were added or removed, so all that updated were indicies

Sean (Sep 04 2020 at 19:40):

I see all 10 args getting accessed in the implementation

Sean (Sep 04 2020 at 19:40):

so maybe a step back -- what's the problem?

Sean (Sep 04 2020 at 19:42):

they all end up getting exercised when put_tree_into_comb_and_export() is called

Sean (Sep 04 2020 at 19:42):

notably, the 'comb' structure had most of them directly saved prior to export

starseeker (Sep 04 2020 at 19:44):

Josh reported an issue with the comb dialog in mged, which I can can confirm

starseeker (Sep 04 2020 at 19:45):

If I open all.g in mged, don't do anything, and hit apply - I get an error from put_comb and the ged_tmp shape is left in the .g file

Sean (Sep 04 2020 at 19:45):

There was an issue before too -- at least one bug was fixed, maybe another introduced?

Sean (Sep 04 2020 at 19:46):

that was a test case I specifically worked on...

Sean (Sep 04 2020 at 19:46):

/me tests

Sean (Sep 04 2020 at 19:46):

is it a tcl error or a libged error or can't tell?

Sean (Sep 04 2020 at 19:49):

okay there is something fishy in the commit history. if I do a log on put_comb.c, it's not listing some of the commits that were made to it before it moved (including the arg reordering)

Sean (Sep 04 2020 at 19:56):

I see now. Subversion apparently has a new git-like feature where its hiding commits that get incorporated via merging. So despite changes that happened on trunk in the r75719-76352 range, those commits were merged to a branch that was merged back to trunk in 76352, so svn is hiding them (-g option shows them). That's messed up...