Thank you for all of the bug-fixes and changes, and the work that went into them! 
Have some more blog posts!
3 Likes
To start with, congratulations on launching the funding campaign, and on so quickly reaching your first milestone! I hope that the campaign continues to go very well indeed.
As to the new release, thank you for it, and for all of the work that went into it! Itās appreciated! (Especially that stuff about monitor resolutions and fullscreen supportā¦ ;))
long time no see, but now Iām back to write some more posts. Hereās the new one!
2 Likes
Iām really glad to read that funding is going so well! And good luck with finding additional developersāI hope that you find excellent ones. 
Blog post! A release announcement this time.
1 Like
Aah, this is excellent news! 
(Well, not the troubles with buildbot; Iām sorry to read of that! Hopefully itās all resolved by this point!)
The inclusion of icon-support is a nice surprise!
I have the new version installed now, I believe, and am excited to try it out!
Iām curious about something regarding the new icon support, if I may: the blog-post indicates that the system will automatically take a 256x256 icon and scale it down to produce other sizesābut will it do so with other sizes, or is it specific to the 256x256 size? If, for example, I donāt provide a 256x256 icon, but do provide a 1024x1024 icon, will it still generate additional sizes?
Yes, it will always pick the next larger size to resize from (except that it will prefer to resize down from an icon with double its size, so it will generate 256x256 from 512x512 even if there is a 300x300 one). So you could even provide a 987x987 file if you were so inclined.
It is recommended that you hand-optimize icons with the lowest sizes (16, 32, 48) because scaling those down from 1024x1024 will probably result in a blurry mess.
Yeah, the buildbot trouble is resolved; we got a brand new Windows buildbot set up, courtesy of the CMU ETC.
Very neat indeed! Thank you for the answer.
And donāt worry, I intend to hand-resize my smaller icons at least! I was mainly just curious as to the behaviour of the new functionality.
Iām glad to read that the buildbot trouble is sorted out, and of the aid from CMU in that.
Iāve just released Panda3D 1.10.4.1 to remedy a critical regression that was found in 1.10.4:
https://www.panda3d.org/download/sdk-1-10-4-1/
Apologies to those who had already downloaded 1.10.4, and thanks to @wolf and @Thaumaturge for finding and reporting the issue quickly!
I would like to use this opportunity to invite people who know Python and use Panda3D to take the time to pick their favourite (or least favourite) part of the Panda3D API and write some simple unit tests (using the py.test framework). Improving the test coverage of our test suite will greatly help reduce the chance of regressions and make Panda3D more stable overall.
Thereās a new release out! The announcement is here:
It has more than 50 bug fixes! It is particularly of note for macOS users since it fixes some critical macOS issues on the latest release of macOS, āCatalinaā.
5 Likes
Ah, congratulations and well done on the new releaseāand thank you for it, and for all your work on it!
1 Like
I subscribed to github, I have this word is associated with sending spam. It seems that this OS is popular as " VISTA" 
On the download page, there is only one version of the third-party tools for macOS and itās not mentioned if itās for the macOS 10.6 or 10.9 SDK (Though I guess itās for 10.9, mentioning it clearly would avoid some issues in the future )
The simple answer is that the thirdparty packages now support both.
All the C thirdparty libraries (ie. the ones that donāt have to be linked with libc++ or libstdc++) are still compiled with 10.6 and continue to work on all the macOS versions after that, so there was no reason to rebuild them for 10.9.
The C++ thirdparty libraries come in two versions: in the ālibā subdirectory are libraries built with libc++ and the 10.9 SDK, and in the ālib-10.6ā subdirectory are versions built with libstdc++ and the 10.6 SDK. When building the 1.10 branch, makepanda is smart enough to grab the ones from the ālib-10.6ā subdirectory instead of the regular ālibā directory when building against an older SDK.
1 Like
Great, itās even simpler like that
New post:
3 Likes
Congratulations on the new release! It looks like it has some neat changes.
I do have a problem, however: it looks like upgrading to 1.10.6, having previously been using 1.10.5, introduces a bug in my game. :/
(Iāve double-checked by reverting back to 1.10.5, and the issue does indeed vanish when I do so.)
My first instinct is to think that itās related to Bulletāitās essentially a collision issue, by the looks of it. However, looking at the commit list for the 1.10 branch, there seems to have been only one commit related to bullet since before 1.10.5 was released. Am I understanding correctly? Or is there somewhere else that I might check?
If I am correct, then thatās not likely the cause, and Iāll have to look further afield for the change that introduced the problemā¦ :/
[edit] (Noting that this is a guess.) Perhaps some change related to the use of mask-filtering, as specified using the ābullet-filter-algorithmā keyword in a PRC fileā¦?
Ah. Thatās a pity.
There is only one Bullet-related change in 1.10.6, which was a misdeclaration of a type in BulletHeightfieldShape. Youāre not using that, right?
There is a change related to the collision system, as outlined in the blog post, but this shouldnāt affect Bullet.
Could you give more information about the bug in question?
Indeed, I spotted that one, and indeed, Iām not using a heightfield-shape.
The setup is sufficiently complex that right now Iām not entirely sure of quite where the system is falling down.
I imagine that I could figure it out through a sufficiently careful examination of the scenario, but I think that the following might be easier:
This was presumably a result of something introduced since 1.10.5. So, today Iām planning on checking out the 1.10 branch to my computer, and then doing a binary-search hunt for the commit that introduced the issue. Hopefully that should uncoverāor at least point toāthe source of the issue!