Hi,
I received a few requests for hardware purchases, that I think are worth
discussing with the project as a whole in order to progress towards
having clear guidelines for what is acceptable and what isn't in terms
of spending Debian money.
Please provide feedback on the proposed decisions -- they are not final
yet.
A. Memory expansion cards for m68k buildds (expected cost: 500 EUR)
===================================================================
Widely quoting from a private mail:
| Debian has an unofficial m68k ("Amiga") port. It became unofficial
| mainly because build daemons were not fast enough to keep up with
| security updates in a timely manner. The port has an active developers
| community (see https://lists.debian.org/debian-68k/). Debian is the only
| Linux distribution with a working m68k port, and m68k is still popular
| among fans of retro computers. It takes little financial effort to keep
| the port running; what is needed is money for small hardware upgrades.
I'm inclined to APPROVE the request, for the following reasons:
- even if the port is unofficial, and is very unlikely to become an
official port again in the future, it benefits from an active
developers community composed of several DDs.
- experience has shown that porting work on one architecture often
benefits other architectures (since similar problems show up
across different architectures).
- the amount of requested money is relatively low.
Obviously, the following conditions would apply:
1) the hardware bought would have to go to buildds used through
debian-ports.org, and/or to porter boxes that are widely open to the
Debian community.
2) if the machine stops being used for Debian-related work, the
hardware must be transferred to another DD.
B. Powerful machine for d-i development (expected cost: 1.5k-2k EUR?)
=====================================================================
A debian-installer developer writes:
> up to now, I've always used my own hardware for Debian work, but I'd
> like that to change slightly due to my work on d-i. I intend to work on
> at least the following topics:
> 1. performing more frequent d-i uploads:
> http://lists.debian.org/debian-boot/2013/10/msg00194.html
> 2. implementing some kind of official images with backported linux
> kernels (and possibly other needed bits from the right suite);
> 3. implementing automated regression testing, so that we can work
> properly on 1., 2., and also on stable uploads; dailies would also
> benefit from that; people from -cd@ (Steve, mostly) would probably
> appreciate it as well.
>
> Some desktop machine with fast disc(s), a bunch of RAM and some CPU
> power would be nice, so that I could play with a bunch of VMs (most
> likely, primary targets will be amd64 and i386, but virtually anything
> qemu can deal with). Some disc space to hold a local (possibly partial)
> mirror would be a plus, since there's plenty of deb/udeb fetching during
> d-i builds and when testing installation.
>
> Do you think something can be arranged on Debian funds to that effect?
> If that looks reasonable, any specific site/vendor I should be looking
> into to come up with some specs that would be nice to have, so that you
> ACK/NACK it? In which case, any upper bound? Or any other ideas? Of
> course the HW can be shipped over to the next person wanting to work
> that much on d-i in case/when I start burning out. (FWIW I don't plan on
> leaving the d-i RM position before jessie is released. ;))
>
> [ Also, I do realize having some nice hardware racked up in some
> datacenter would be nice for testing purposes, but until automated
> regression testing is implemented, one needs to rely on clicking and
> typing into a VM, so as to debug/develop some framework to perform
> automated testing. A datacenter-hosted machine would also not help with
> the “preparing an upload” side, which still needs some trusted, local
> machine IMVHO. ]
I'm inclined to APPROVE the request, for the following reasons:
- the machine would be primarly (only?) used for working on Debian
- the specifics of the tasks justify hardware hosted locally (VNC to
a remote machine is possible of course, but latency makes it quite
inefficient to do testing that way)
Of course, as quoted in the original mail, if the DD were to stop their
involvement in d-i development before the machine reaches its end-of-life,
the machine would have to be shipped to someone else.
C. Laptop for developer (expected cost: 1k-1.5k EUR?)
=====================================================
A DD is asking for help to buy a new laptop. He maintains or participates in
the maintenance of a few medium-to-large packages. His only mobile computer is
an Atom-based netbook with a rather small screen, which is not powerful enough
to do packaging work (he also has a desktop computer). He describes himself as
a "first world middle class person", is currently a student, and cannot afford
to spend money on hardware.
I'm not sure of what to do:
- Debian cannot afford to buy hardware for every DD
- But many DDs don't need Debian to buy hardware for them
- This machine is more general-purpose, not specific to Debian work than in (B)
So far, I answered with the following question:
> How much do you honestly think that this purchase will increase your
> Debian productivity?
But I would welcome other ideas of criterias to apply here.
Lucas
Attachment:
signature.asc
Description: Digital signature