On Sep 26, 2013, at 5:44 PM, "Nikola M." <[email protected]> wrote:
> On 09/26/13 05:02 PM, Garrett D'Amore wrote: >> >> On Sep 26, 2013, at 6:59 AM, DavidHalko <[email protected]> wrote: >> >>> An attempt at migration of capability before EOF seems to be reasonable >>> planning. >> >> That would be true *if* we knew of any active use of them. I don't. >> > I see one of the strengths of Illumos of supporting legacy hardware with > legacy drivers, Uh, not really. We don't support a lot of ancient hardware anymore… ISA, 386, 486s, UltraSPARC-I, and even some "newer" SPARC based laptops are no longer supported. Because nobody is using them. Supporting "old stuff" hurts people trying to do newer things; if that's what you're looking for (as a value in itself), pick up NetBSD. That said, the stable DDI *is* a value. But that's orthogonal to what we're talking about here (mostly). > most like Linux can support all older hardware, that was once contemporary, > that some other platforms ditched supporting long time ago. > While taking the most out of old hardware is nothing wrong, and one of the > strength of the platform, it is hard to choose what not to support anymore. > > But, if those classes of drivers and their kernel support for it, make some > trouble building, maintaining or something, then no one can blame nor > actually stop developers from ditching it. > > Q: I would like to know how high quality would then be reservation of an such > old kernel code (where would it reside in some illumos repository, once > removed) and how hard would be for someone wanting suddenly in a modern > illumos kernel, to add support to the kernel for that class of devices (PC > Card, 16 bit PCMCIA), if he/she want for it's hardware? > Would it be impossible to add it or one would be needing to use older illumos > and not using newer illumos to make them work? > Or it could be done in such a way that "legacy kernel interfaces" could be > extracted and made available together with removed driver class support in a > way that they could be easily added if needed in later brand new illumos? > > Personally, I have IDE over PC Card, Fax/Modem together with LAN and Wireess > Lan card, lying around, not actually using it. > I have PC Card slot in a latop (32bit I suppose), that I actually did not > use, since I think my PC Card slot is broken or something. > > I would like to point out, that in the line of Laptops sold around here at > shops, > there were evident sharp decline of a number of laptop models that used to > provide even PCI ExpressCard in various sizes. They (importers) just at once > stopped offering widely laptops with these external expansion options. > There were wide range of laptops with these slots several years ago (some > with PCCard, some with ExpressCard, but now (last 4-5years) it is very hard > to find consumer grade laptops that come with it. > I would for now actually dislike to see that ExpressCard support disappear > since it is actually still contemporary way of adding external high speed > devices (external graphics etc). > (Someone talking about adding Thunderbolt (2) and USB3 support?) There are reasons that ExpressCard died. Those reasons are mostly USB. :-) > > Related to that, About Smart Card readers, I would be strongly against > ditching Smart card reader drivers, like Solaris11 ditched them (I know they > are not related with PCCard/PCMCIA and that they differ). Smart cards are > used on modern credit cards, personal ID cards and e-banking cards for > identification, containing signing certificates and for network > identification while logging in. > I would not like for that support for Smart Card to go away any time soon > from illumos. We need that in any application of illumos desktop/headless > implementation for identification in foreseeable future. > > How hard it is to have older stuff supported anyway - Does it make much > trouble? Basically, you're advocating keeping the crufty support while admitting you have such hardware lying around, but don't use it. :-) I think that's a good example of why we should not be making decisions to keep this stuff if it costs *anything* to keep it. It turns out that supporting ancient hardware is *not* free… breakage occurs. And in the PCMCIA case, it means we have to support older interfaces, which can limit some other optimizations. In the case of PCMCIA, after the removal, it will be non-trivial to add it back, because the kernel will fork over time. That's the *motivation*, actually. It's time to kill this ancient stuff. - Garrett ------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com
