Ok … responding to this part first:
3) I don’t see the need of this. I think our CI servers are now setup nicely.
We have build-times of about 8-10 minutes. I don’t see a need to jump through
all the hoops of having to maintain our own infrastructure. Yes, the Ant build
doesn’t run on Infra, but I don’t see this as a problem of Infra, but more of
the Ant build. The Maven build works nicely and I currently can’t think of
anything that’s not possible at the moment. We now even have automated UI tests
in Browsers and whenever I needed anything, I got it in the end. Yes it did
usually take some time and convincing, but in the end things turned out exactly
as we needed them.
4) I don’t see this as a valid argument towards requiring our own Jira … ok …
it is somewhat sluggish every now and then, but for example, being a TAC judge
I like the option to search for a user’s Jira activity over several projects.
If TLPs started having their own JIRAs, things would get a lot harder.
So now to some general thoughts:
- If we started this, probably a lot of the big projects would start getting a
lot of funds. Funds that no longer go to the foundation, reducing the funds
shared upon projects without their own funds. We have 3xx TLPs, but I doubt
most of them would receive the funding they would need. So, we would have a
hand full of rich projects and a lot of poor ones. I don’t think that’s the
spirit of the Foundation. What about Apache Commons? Probably one of the
most-used project, but probably also not one of the sexiest ones for getting
funds? What about the Incubator? It’s one of the most important Projects of
Apache, but it doesn’t have that outside marketing value.
- Having companies pay for individual Projects sort of produces a bad feeling
in my gut. Even if they don’t say so, they still could and probably would
expect something for their generous founds. I don’t want to be in the situation
of having to implement something a company wants, because otherwise they
threaten to cut funding.
- I think we suck at running infrastructure … at least compared to having a
team of skilled people watching for Infra 24/7. What happens in the time lights
are out in the US, if a service goes down just before Christmas? I don’t want
to rely on volunteers to run our systems. That’s also the reason why Infra
people are the only paid contractors at the ASF.
- There is a huge number of things you must deal with when running your own
infrastructure. A lot of the things are quite restrictive on Apache Infra, but
it’s not because they want to make our life harder, it’s because of damage that
had been done in the past. We would have to re-learn all those lessons already
learnt by Infra. Setting up a server is super-easy. Setting one up that’s not
hackable in a handful of minutes by a Script-Kiddie is super-hard.
So, if it’s just that one VM you are looking for, I don’t have any objections
as I don’t see any difference to what we are doing now. But I don’t want to
have an essential part of our Projects infrastructure outside of Infra.
Chris
Am 10.02.17, 22:46 schrieb "Alex Harui" <aha...@adobe.com>:
My reasons for spending time on this are several.
1) The ASF is growing and so are expenses. Expenses rarely change based
on the economy, but donations can. If a time ever comes to discuss
cutting expenses, I want to protect Flex by being able to make a case that
we don't cost the ASF any money and in fact, help.
2) We should be donating money to the ASF, but how many of you do? I
don't myself. And would it be easier to justify if the money went
directly to Flex? It would for me.
3) I am currently paying for one of our CI servers. If we could get
directed donations to pay for it, we might be able to upgrade to a faster
server. I would personally donate more since I would get a tax break on
the donation. And anyone who wants to pitch in can help and at least in
the US, get a tax break.
4) There are certain resources we share in the ASF like JIRA that are not,
IMO, optimally set up for us. We can't create custom JIRA fields, for
example. And more than one person has tripped over the Infra-centric
buttons on each JIRA issue.
5) I, and I think several board members, want to understand if handing off
more server responsibilities to the project would scale to other projects
and help the bottom line or hurt it.
