Understanding the Qtplasmac update process

beefy wrote:

So the manual tells us to use the "buildbot" method. Can someone give me a rough idea in layperson terms what buildbot actually is, and why we use that for minor changes to Qtplasmac.

buildbot.linuxcnc.org/buildbot/
buildbot.linuxcnc.org/
Buildbot takes some time to build new "builds" of LinuxCNC as it has to build a lot of stuff with many versions supported on many Linux distros and processor architectures, just have a look at the first link.
Usually a day or two, but if any errors are found during the build process, it takes more till the errors are fixed.

I think the short answer is the buildbot may be stuck, it seems to happen from time to time.  I tried to pull down the updated package too and it was still version 283.

In the simplest terms, QtPlasmaC is a GUI that's part of the LinuxCNC project.

You have a few main methods of using LinuxCNC (the docs are not included in this discussion):
1. Pre-Compiled Package install (buildbot) - this is what you have, it utilizes pre-compiled .deb files to get LinuxCNC onto your computer.
2. Run In Place (RIP) - this requires cloning the github project, and then compiling the project on your computer to be able to use it.  
3. Compile the packages yourself from the cloned github project and install them.

The end result is essentially the same as far as how they are used.  What is different is how accessible the files are for development.  Most of the developers here use #2, and most general users here use #1.  

For option 1, these pre-compiled packages are made via the BuildBot.  Any time a change is made to the project, the buildbot gets busy compiling the .deb files for each of the systems listed here:  buildbot2.highlab.com/buildbot/#/  (this is for the new "secret" buildbot)  So, when we make a small change to QtPlasmaC, the buildbot compiles the entire project.  This creates a new build, and is essentially your new "version" of LinuxCNC.  The worst part of this option is the wait for the build.

One appeal in using Option #2 as a non-developer if you can handle all of the commands (which are actually minimal after the initial compile) is that the changes that are pushed to the github project are available immediately for you to pull to your system, and compile locally.  Generally speaking, most users are turned off by this because it may take longer (the changes I pushed would have been instant though, as they didnt require any compilation), sometimes goes awry, and requires launching linuxcnc from the terminal (or making a purpose built icon to do it for you).  You dont get the nice menu list (windows-like) that most are used to for programs.  There's other things too, but I dont want to bog this down further than I probably already have

#3 is an option too, but takes even longer, and more steps to have an updated machine.

Hopefully that helped more than it hurt.

snowgoer540 wrote:

#3 is an option too, but takes even longer, and more steps to have an updated machine.

Two things to add here,
-time it takes to compile depends on processor speed and core count, faster and more cores=faster compile
-time also is very dependent on if compiling documentation, it is much faster without documents
Last time i compiled on a quad core laptop it took probably 5-8 minutes without docs
On an 8 core Ryzen 7 3700X took about 2 minutes, probably, i was distracted
--
And another thing while we are at it, this is for anyone able to add to compiling LinuxCNC howto's
Please add a line on how to compile without the documentation, it does cause failures to compile very often and also takes a long time, like really long time to compile with them.

rodw wrote:

Tommy, the available parameters are full documented in the building Linuxcnc docs under heading 3.2.1 near here
linuxcnc.org/docs/2.9/html/code/building...ding_debian_packages
 

Oh good, thank you.