Hi, folks, I had read the long discussion history about mmj2 installation on github:
- https://github.com/digama0/mmj2/issues/14 - https://github.com/digama0/mmj2/issues/21 - https://github.com/digama0/mmj2/issues/36 - https://github.com/digama0/mmj2/issues/37 - https://github.com/digama0/mmj2/issues/39 Lots of the questions had already been solved, some still haunted there. Especially in Issue #39, @digama0, @david_a_wheeler and @cwitty discussed the problems and the engineering complexity for future fixes. After considering all the above in an hour, I don't think the installer in my previous mail is a good solution, it is only one possible option. I respect all the contributors' works and understand the status of open source development. We had already improved a lot, but still did not achieve the goal For me, there are two sets of engineering problems beneath domain problems, maybe it is biased on my preference and experience. 1. homework: - 1.1 the build system: ant is OK, nowaday developers use gradle and others, the tool chain of Java evolved a lot. - 1.2 lack of test cases stops people outside to contribute further. 2. application problems and solutions - 2.1 multiple OS support: at least Windows/Linux/MacOS - 2.2 multiple Java environment support: Java 8 and aftermath, esp. the LTS versions - 2.3 a well-designed convention and workflow on metamath/mmj2 develop: where do we put the database and personal works, etc. Above these pure engineering problems, it is the problems of metamath and math, the beneath and above two parts are orthogonal. It means a developer can focus on engineering problems without introducing too many domain problems. As any open source projects, the engagement, trust and time are real problems. I used to be a Java developer for years and I can contribute some. I also propose a way to solve these: - migrate the build system to a new generation builder - give sanity tests on set.mm and other databases - give more test cases on basic functionals - once tests are ready, we can integrate a CI to check on different OS and JVM By this way, we can define our engineering problems very clearly, all the problems are surfaced on one page with different status on different platforms, and then we can fix them one by one. @digama0, @david_a_wheeler and @cwitt, what do you think about the proposal? Mingli On Sun, Feb 13, 2022 at 5:08 AM Mingli Yuan <[email protected]> wrote: > Hi, folks, > > After 1 day of investigation and 1 day of work, I set up a demo repo which > can generate installers to ease mmj2 installation. So far only the Mac > version *.dmg was tested. > > The demo upgraded to a higher JDK version, and then generated installers > by jpackage which was shipped by JDK after version 14. > > The main point is > - JRE was bundled with the installer together to ease Java installations > - use hidden user directories to store default settings and read only > database to eliminate initial configurations > - Runnable after installation without any configurations > > Since it is a quick hack and demo, the code was a little bit dirty. > If you think it is helpful, I will put more energy on it. > > Comments and ideas are welcome. > > https://github.com/mountain/mmj2e > > Mingli > -- You received this message because you are subscribed to the Google Groups "Metamath" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/metamath/CAGDRuAjO8cvhb0u3P3EAnEp9qMfPEJJVZBzryksWUeRYT0unhg%40mail.gmail.com.
