[Amforth] Followup: AmForth Project open for adoption

[Erich W. <ew....@na...>]

Sorry, hit the button quite prematurely :)


Dear AmForthers,

thank you so much for your kind words!

So far a person behind the email address "in...@un..." has
offered to help with AmForth. I have suggested to make an
appearance on the list.

Martin Nicholas has suggested to forward the original email to
comp.lang.forth. I have asked Bernd Paysan (of gforth fame) to
forward the original announcement. So the news it might reach a
greater group.

Brian Navarette has asked, whether the code of my "small" fork
is available. Yes it is, see below.

One of the things I did, but never announced is this: I
converted the subversion repository to git in two variants:

In this repository the release tree is spread out like in the
original subversion repository:
> https://git.sr.ht/~amforth/code-tree

And in this repository the releases have been converted to git
branches:
> https://git.sr.ht/~amforth/code

The account 'amforth' on sourcehut.org is a paid account. I'm
happy to provide funding for this account for the foreseeable
future and pass it on later. Carsten Strotmann has reserved the
domain amforth.net. currently it points to the sourceforge page,
if I am not mistaken.
> https://www.amforth.net/

So these are things to play with.

The web interface for Sourcehut.org works completely without
JavaScript, which is why I have chosen them and migrated all my
repositories. On top, changes on tickets and code can be coupled
to email. So changes can be managed with email workflows.
However, I have not understood all the neccessary details. I
don't remember just now, whether I configured a new mailing
list, probably not. Because I never really tested the workflow.

But these are all puzzle pieces to play with.

I admit that I have hesitated with the mailing list move as
well. I am a little concerned that this move might break the
small community. However, I would be more than happy to be
proven wrong.


So where is the above mentioned fork of AmForth Version 5.0? You
can find it in this repository:
> https://git.sr.ht/~ew/hbv3_am50forth
The corresponding project featuring the rs485 connected
controllers is growing at
> https://git.sr.ht/~ew/hausbus-v3
The hardware I use is described there as well
> https://git.sr.ht/~ew/TheStack

The second test is running happily and blinking its LED. 544000
seconds uptime without a hitch, so it's about halfway of what I
want to see (10^6 seconds or 12 days). I have started to
document this journey, and it should turn into English text
along the way.


Conclusions:
- Though I have stepped down, I am still the one with the keys,
  the janitor, so to speak.
- I don't think I am going to drop off the planet soon. Or at
  least this risk is not much greater than it was before.

What can you do?
- I would suggest that 3 people receive the credentials for
  sourceforge and sourcehut, just to reduce the bus factor. I
  suggest Carsten Strotmann, Tristan Williams and Martin
  Nicholas, but that is just me. Speak out, please.
- I still have a half baken document, how to make a release. So
  I should just send this as is, or add it to the repositories.
- everyone can make up their mind, if they prefer to move to git
  and sourceforge.
- whoever has more insight to the puzzle pieces at sourcehut can
  try to play with a email based workflow. I strongly suggest to
  create a public bug ticket instance, I percieved this as a
  significant lack in the past.

I would like to read your thoughts on the list. I have no idea,
how many there are reading. The number of subscribers is
probably not a good estimate. So, speak up, make yourself heard.
Thanks!

Cheers,
Erich


Erich Wälde <ew....@na...> writes:

> Dear AmForthers,
>
> I am herewith stepping down from the maintainer role of AmForth. For details,
> read on. If anyone is interested to take over, get in touch with me.
>
>
> In 2020 I received the logins of amforth.sourceforge.net, basically because I
> was lucky enough to have met Matthias personally a few times. At that time I
> had a lot of ideas on how to proceed. And while I managed to create an
> official release, there are a few obstacles in this path.
>
> First and foremost I am facing a health issue. It is subtle, but it
> seriously limits, what I can do. I still have to make several difficult
> decisions regarding my daily life. I have started to decrease the number of
> things on my list by cancelling items. I have to accept the fact, that I'm not
> in a position any more to advance the AmForth project in a meaningful way.
>
> Secondly, AmForth has become complex over time. Matthias added support for
> three more target platforms (msp430, arm, riscv32). Allthough access to these
> is easily achievable, I use only avr. And I use it less these days.
>
> Thirdly, AmForths tools are depending heavily on python code, a language I
> consider myself illiterate in. I have written a few small perl scripts around
> AmForth to serve my needs. I heavily depend on those and on a Makefile.
>
> Add the fact, that in 2020 I spent countless hours to port my data acquisition
> stuff at home from amforth 4.6 to 6.9 and it just did not become stable. To
> this day I still have no clue, why the thing hangs itself after some time,
> sometimes hours, sometimes several days. In other words: unusable.
>
>
> Step back: what would I have done, if I didn't know Matthias, and the project
> would just have become silent? Simplify. Simplify heavily.
>
> Very recently I have made a fork of AmForth release 5.0. That version is
> before support for msp430 was officially added (5.5). That version happens to
> compile with avra rather than wine/avrasm2.exe. Along the way I found, that
> avra has seen new releases, which add support for my beloved atmega644p and
> lots of fixes, which is nice! This removes the dependancy from wine and allows
> for smaller systems for development.
>
> So I have picked up my data acquisition project again with the fork mentioned
> above. Any Interrupt Service Routines are written in assembly to avoid the
> thing that I uncovered in 2017, namely a race condition 1 bit wide and 1
> instruction cycle long. I pick missing bits and pieces from later releases. I
> would like to add a few features regarding sensors with different needs. A
> first experiment has run more than 10^6s (12d) without any failure. So I am
> moderately optimistic to continue along this simplified path.
>
>
> Thanks to all, who have answered the list, contributed code, ideas,
> documentation in one form or other. It has been an interesting experience.
> And should you still care to listen: if you have one or a few more important
> plans, do not delay them, you might be unable to pursue them later.
>
> Happy hacking, and use the Forth!
>
> Cheers
> Erich

-- 
May the Forth be with you ...