Thanks for the next great release of KiCad!
Two questions, from a packagers point of view:
Is there already a changelog available, listing user visible changes.
As 5.0.1 is just a bugfix release, it should be safe to use it with the libraries from 5.0.0, correct? Especially the 3d packages are a huge download (not everyone has an uncapped and fast internet connection), so not forcing (just recommending) an update would be welcome.
That is something you need to ask the devs over at the mailing list.
Yes
Thats what delta packages are there for. dfn (fedora, redhat, …) supports it for years now and i would guess apt (debian, ubuntu, mint, …) should also support it. (Windows and maybe also mac are a different story)
Another option is to use git itself. This requires the libs to be separate packages such that the user can select not to install them via the normal packages. (Sadly again maybe not done this way for all platforms. Windows for example does not offer such a package)
You can just go to GitHub and manually select recently changed models of interest.
It is a pity that there is no rsync option.
I’m pretty sure that If you clone the git repo, and do a pull, you will get the diff and not download the whole thing again.
Thanks, will do.
So does openSUSE (these are the guys who invented it …), but only for regular releases. For TW and addon repos, it is to much churn.
Although for the kicad-packages3D rpm, it is not really a big gain - after 12 minutes of processing (i7-4790), you end up with a 200 MByte drpm instead of the regular 320 MByte rpm.
This would be an option if it were better integrated in the library manager - i.e. already added with a download/refresh button. Or just a “Download packages for all footprints used in the current layout.” button.
Would it be easier if there is a dialog inside kicad? Yes it would.
But saying this is not an option without it is simply wrong on so many levels.
You really do not need to include everything into the same toolset. Git at this level is really not hard to understand. (This sort of thing is easily doable within any git gui)
An update has been posted. The next release is now very near. (Wayne wants to have it done by tomorrow)
https://lists.launchpad.net/kicad-developers/msg37914.html
Nor do you want to with finite developer power available. Beter to address issues/functionality which can not be solved with external tools.
As I said, I a posting here with my packagers hat on. I am completely fine with using git, but I do not expect everyone else to be.
There are two possibilities here - either require the datasets or let the user handle it. If the datasets are not available, the relevant function (PCBnew, EEschema, 3d Viewer) will not work as expected (no componen/part to place, no component shown). As there is no error/warning message shown, a possible first time user will get frustrated …
For the symbols and footprints (which are separate packages for us) we have chosen to require them, but for the 3d packages weighting between size and need resulted in keeping it optional.
Of course using git is an option from a “is it working perspective” - but not as a default choice. If library management were so easy, you would not have to devote any FAQ entries to it. It is not only about using git, but also the awareness it is available.
I do no ask for a git replacement, but for a well integrated way to use it. Even a message in the library dialog would help significantly. If you push away first time users, it gets hard to gain new developer resources.
Not all developer resources are equal. This could be a task for someone who is relatively new to KiCad development. For sure it is not something critical, but what is wrong with acknowledging it is a weak spot?
I have contributed to the KiCad documentation build system in the recent past, but getting these kind of answers let me reconsider any further involvement.
This is all already there with the standard packages. I merely suggested an alternative as you wanted a way to download only needed information. Git is simply king in this regard. Everybody who wants an easy solution just has to deal with the fact that it will be more expensive on the download volume side. (Everything in technology is simply a tradeoff. In this case ease of use vs. download volume)
People often say that sort of thing, but I’ve never seen any evidence for it. I don’t think the two things are in any way linked.
The “if you don’t play my game I will take my toys away” approach never gets any traction.
Even if we all agree that seamless git integration was a vital and necessary feature, you can’t just wave a magic wand and make it happen.
The Linux packaging is quite a mess, which Windows users don’t suffer, but all platforms suffer from a lack of integration with libraries. It should be a lot easier to install/update/remove libraries.
Personally I would like to see a tools API that allows any VCS to be added via a plugin, but it’s far from a show stopper.
Why do you think packaging KiCad or the libraries is being a mess on Linux? Contrary to Windows every Linux distribution has a package management system and packaging the stuff is a almost automated thing. I don’t see any mess here. Understanding the package management is one thing, the most important thing is the human resource on this part.
It might be a bit confusing as there is not one standard of how the packages are named and split. Some distros use one large package for everything while others have the libs split away into separate packages. (These later distros sometimes supply a meta package for adding all library packages with one command. This results in another inconsistency.)
Being slightly off topic, I’ve found that a lot of times on this forum (as this is the only forum I regularly participate where members are from all around the world) that two persons can express them self completely differently due to different language, cultural background and professional experience all the while meaning the same thing.
Back on topic, I agree with you that even a little bit of VCS (git or in general) integration support would go a long way especially for handling the libraries, but I haven’t got a clue where to begin with. And even if I had an idea what to implement, that would still not be enough as C++ is not really my forte. Thus I dable a little bit with python and try to expand the KiCad ecosystem with action plugins.
As for the documentation, I also agree with you, but I’ve found out that before publishing anything, it has to be tested, as it can do more harm than good (a big thanks goes @bobc). Thus I am testing the documentation how to integrate git with KiCad library management in my local KiCad usergroup in my mother lanugage, and when it will be refined, I’ll try to include it in official documentation
In the same way, two users may appear to be speaking of the same thing, or may appear to be in agreement on some point, when in fact they are not. Posting screen shots or the application’s native files often helps to resolve both types of error. The discrepancy may not originate with language or culture - it may be due to differences between hardware engineers and software engineers; developers versus “ordinary” users; coming to the discipline of PCB layout from mechanical versus electrical training; previous experience with some other EDA tool; or even chronological age.
When a multi-national, multi-cultural forum (such as this one) is successful, the users have learned to allow a wide “tolerance band” for language usage, and to consider the possible alternative meanings for a post. I am asking the Forum members to forgive me for the times, both in the past and in the future, when my vocabulary, English language structure, or logic are simply too complex, or ambiguous, or misleading. It is not intentional.
That is an excellent start. Especially if your local KiCad usergroup has not had any involvement with developing either the KiCAD program, or the documentation. The next step, of course, is to test any translation on a similarly uninvolved group.
Dale
P.S. - I looked on your user profile for a clue to what your first language might be. You have not put any information in the “About Me” section.
It’s possible that you will never see 5.0.1. Instead 5.0.2 will be released within a week. For the reason see https://lists.launchpad.net/kicad-developers/msg37932.html and forward.
1 KiCad week is about 8 human weeks. Therefore the prediction above was correct. See https://lists.launchpad.net/kicad-developers/msg38454.html.
God, if I only have 1 month to live let it be in NBA final two minute time? 
Apology to those that don’t get the sports reference.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
