Re: [jdt-dev] Telling GitHub to rebuild, rebase, ...

[Stephan Herrmann <stephan.herrmann@xxxxxxxxx>]

Am 12.09.22 um 10:11 schrieb Christoph Läubrich:
> So I can only assume that this "meta-question" was not a real question but more 
> a "joke" 

Actually, there *is* more to it than a joke (despite the smileys in my post). I 
might continue down the alley "I can only search for things which I expect to 
exist", but still that's not the real reason why I asked the meta question.

Let me pick out just three paragraphs from the very mail to which Christoph 
responded:

> > > Am 11.09.22 um 15:38 schrieb Stephan Herrmann:
> > > > Perhaps the discussion got derailed because people saw "rules" ("thou shalt 
> > > > do X") and were quick to reject anything limiting their freedom.
> > > >
> > > > What you mention concerns the pursuit of helpful information ("how can I do 
> > > > Y?"), which doesn't require us to agree on any discipline.
> > > >
> > > > Both kinds of discussions _could_ converge into a shared, living document 
> > > > "how do we do Z?". And that document should be posted right on the front 
> > > > door to JDT.

Three simple ways to ask different questions.

Now, here comes my compiler engineer attitude, in which each word can make an 
important difference (you need to be very, very careful with words when reading 
JLS, as well - of course - when you try to translate that into java).

I did NOT ask:
  "how can I contribute to CONTRIBUTING.md?"
This is the question Christoph answered, but I asked a different question:
  "how does one contribute to CONTRIBUTING.md?"

If you match that against the three paragraphs above (from the same mail), 
you'll see that this best matches to the third paragraph, where I speak of 
blending rules and technical guidance into a common document of - let's call it: 
shared habits.

So now that we all are aware that it is *possible* to edit files directly in 
GitHub, there could still be reasons why we might not *want* to do that. E.g., I 
have been told that editing files in GitLab (yea, different tool) using the 
Browser can severely screw up the encoding or line-endings or such. DONT.

But the thing that really got me thinking: do we want to think of our git repos 
as wikis? wiki-wiki is the technology for quickly editing a shared hypertext 
document. Quickly! The JDT git repositories, however, should be modified only 
with great care and reviews and all that. So the question I was trying to ask, is:

  "Do we want our git repo to be edited in wiki-style using a web browser?"

When only the top-level jdt-git is treated that way, *perhaps* this is fine - 
pending agreement by the team. But if we capture contributing-hints in each 
component's git repo, this would mean that our git repo history will be a chatty 
mix of wiki edits and code changes, right? Do we want that?

If browser-editing CONTRIBUTING.md is OK, perhaps we can do it also to other 
files? Perhaps just fix a typo in some messages.properties text file? Or, hey, 
that code formatting looks weird, let me just quickly fix that in the browser. 
Save. Done.

And finally, when I asked: "Is a PR needed for that?"  -- please imagine that 
this question could perhaps mean: do we *want* PRs in order to apply some 
quality control also to text edits?

I'm just asking questions. Perhaps everybody already knows all the answers, but 
I'm not sure that those will be the same answers across the group. That's why 
I'm more and more reluctant to use the term "team".

best,
Stephan