Mapping SVN usernames to git

[Dirk Stöcker]

Marcello Perathoner wrote,

> Seriously. No need to be overly defensive. I was not questioning your 
> qualification, but your attitude, which might put willing people off.

I always hear might, could, will and whatever. I told you that I have 
more than 30 years experience. I don't need to rely on "might" anymore. 
I'm pretty sure.

> Here we go again... The point was that using git{hub|lab} etc. there'd 
> be many people who could help you with reviewing pull requests, 
> testing, commenting, signing them off, etc. while there's not many 
> people that can help you as things stand today because of your 
> idiosyncratic system.

No. There will NOT be more people. I know that from other projects. I 
have GitHub projects. I had SourceForge projects (and probably still 
have maybe). And experience shows that the tooling is not important at 
all.

The only important point is whether there is a interesting task involved 
or not. And JOSM is a very mature software. The main work for JOSM is 
fixing obscure bugs. There are only VERY few people worldwide who 
consider that an interesting task. Most will only come by to fix a 
single bug which they encounter themselves.

I myself came to JOSM because it was buggy and I wanted to get rid of 
the bugs. And then there were some very interesting development tasks. 
Nowadays JOSM would not attract me. Today I would only be a one-bug-fix 
contributor. But it attracts people like Taylor which I'm very happy 
about.

> Retooling for git would even save lots of time for the same reasons as 
> above. Everybody does and knows git, but nobody uses SVN any more. The 
> only reason for clinging to SVN is maintainer job security.

No. Like many younger developers you assume the tools are important. 
They are not. For an OpenSource project motivation is everything. People 
who are motivated by fancy tooling will soon be bored and go away. So 
any new tooling will bring a spike of contribution for maybe 1-2 months 
and that's it. Afterwards situation is the same. Only now I need to care 
for the new and thus buggy tooling instead of the older well-worn stuff. 
And people who aren't able to use SVN 100% and aren't able to find the 
JOSM GitHub mirror for sure aren't able to find and fix any obscure 
bugs, so I don't care much if the are attracted to JOSM or not.

See the plugins. We have github-hosted plugins instead of SVN hosted 
plugins, because "authors will care". Surprise. There is NO DIFFERENCE 
to before, only now SVN and GitHub must be supported. BTW, to come back 
to the topic, the reason to move plugins to git is to get away from that 
duality and only have one technology where authors abandon their plugins 
and not multiple.

> Of course I may be wrong. But in this case it is very improbable. If 
> you were right, then all projects who switched to git{hub|lab} in all 
> these years would be dying by now. github would have filed for 
> bankruptcy instead of being acquired by one of the biggest software 
> companies.

Where did I say that? You try to convince me that GitHub would make 
situation better that it is now. I simply tell you that it will not 
improve anything important. It will only break a 100% OSS toolchain into 
a toolchain owned by a company which is know to exploit others 
shamelessly.

> Stackoverflow tells us that 96.65% of professional coders use git while 
> 5.96% use SVN.

I'm pretty sure that statics are wrong, as they simply count the numbers 
of repositories. Now JOSM alone has 2 SVN repos, but probably about 50 
Git repos. Git is overrepresented.

If I think about source code repositories for the packaged software in 
openSUSE I'd say Git to SVN is approx. 60/30%.

And I see that I get old and start to sound like my own grandfather: 
It's sometimes sometimes so hard to discuss with you youngsters. You 
assume you know everything because you can operate the newest fancy 
whatever. The next fancy new whatever will be KI and KI programming and 
that will be the everything and the old ways must be thrown away. And 
then the limitations pop up and some years later, when probably 5 
different fancy new things came the technology will be mature enough to 
actually have found it's niche and be helpful. Maybe a short view into 
the future: KI will not solve the task to find and fix obscure bugs. 
That still will have to do a limited number of people worldwide.

For Freedom In Peace
-- 
http://www.dstoecker.eu/ (PGP key available)