I should perhaps keep quiet since I'm not capable of taking on this task myself. However, I think this point needs to be considered even if future Alpha support is still to be frozen and dropped:
The Microsoft-Intel steamroller has literally flattened the microcomputer field into one of dangerous uniformity. Alphas are one of the few diverse species lines left standing, and Linux is now the only viable OS that can carry them into the future for any length of time. True, OpenVMS is still available and supported at least half-heartedly by H-P, but I think the current 8.3 release will turn out to be the last for the Alpha and it is already light years behind Linux in terms of networking tools. True 64 Unix is already dead by virtue of having become unavailable, though in theory still supported for existing users. In spite of this, there are a great many Alphas out there, and not all of them are slow power hogs. The DS10 and similar models will be readily available for years and can perform very well as firewalls, bastions, routers, and even web servers. (I have a web server running on one right now.)
As has already been pointed out, the diversity of the Alpha hardware and machine code instruction set is a significant security advantage especially for a firewall, mail host, or web server where attacks are more likely to occur. This can help accrue benefits to Linux as a whole. As the active market share of Linux installations continues to increase, which seems likely to happen, hackers will turn more focus toward "breaking" Linux security. The existence of diverse hardware with the same operating system and management tools is a major advantage against this problem, and I for one will hate to see the Alpha go. What can take its place? Now that Apple has moved to Intel style processors, there seems to be no alternative line of evolution left.
On Fri, Aug 22, 2008 at 4:28 AM, Paul Slootman <email@example.com>
On Thu 21 Aug 2008, Steve Langasek wrote:I still use an alpha as firewall / mailserver / DNS server,
> Are there other developers still actively using alpha who are willing to do
> the work to maintain it? Or perhaps a more important question: does anyone
> foresee themselves still using alpha three years from now (1.5 years of
> lenny as stable, + 1 year of security support as oldstable)?
taking the way of security through obscure hardware :-)
Any stack overflows or whatever will probably not have any example
exploit code available for alpha's, as hardly anyone knows any
significant amount of alpha machine code...
That said, I'm in agreement with you about how alpha is becoming
obsolete, which I find a bit sad somehow; it had great potential.
In my opinion it's safe to say that certainly three years from now,
alpha can be considered irrelevant, and continued support will indeed be
a waste of resources. If I still want my alpha firewall (assuming it
survives that long) I could build any packages myself, and I expect that
most people using alphas at that time would be in the same position.