Re: Filter design for nsmux

[Sergiu Ivanov]

Hello,

On Thu, May 21, 2009 at 7:46 PM,  <olafBuddenhagen@gmx.net> wrote:
> On Tue, May 19, 2009 at 08:07:37AM +0300, Sergiu Ivanov wrote:
>> On Fri, May 15, 2009 at 9:13 PM,  <olafBuddenhagen@gmx.net> wrote:
>
>> > The real problem is organisational: changing an existing interface
>> > is a very fundamental decision, and the current maintainer situation
>> > regarding the Hurd means that there is nobody to approve such
>> > fundamental changes... It would be very hard ever to merge it into
>> > the mainline :-(
>>
>> Hm... Does this mean that we are somehow discouraged to do any changes
>> to things that already exist in the Hurd?..
>
> Well, I don't really want to discourage it... Just pointing out that
> it's problematic :-(

I see... However, I hope that the idea with O_NOTRANS special
functionality will work and we will not have the necessity to do any
modifications to anything already existing.

>> I'm afraid I fail to see in which way Zheng's changes are less
>> fundamental than the changes we are inclined to do to an existing
>> interface :-)
>
> His work extends existing functionality, but doesn't change the
> interfaces. It this sense, it's a less fundamental modification...

I see...

> Also, you might have noticed that none of his work got merged to the
> mainline so far -- not even the relatively simple patches :-(

I thought I've simply missed the occasion when the patches had been
merged...

>> A fundamental (probably) question: when you say ``translator stacking
>> framework'', do you refer to the existing mechanism of stacking
>> translators or to some future possibility? (the one about fast
>> translator startup or something like this, probably?)
>
> There is no "framework" part nor even a special mechanism for stacking
> translators in the ordinary fashion, so obviously I'm not talking about
> that. I mean a framework for optimizing translator stacking, i.e. the
> "object mobility" stuff Fredrik is working on.

This is clear. I'm setting sail to read your discussion right now.
(Unfourtunately, I had very little time to read the discussion
recently).

>> > (Actually, there are still two processes -- but the functionality of
>> > one of them is migrated over to the other, so one becomes an empty
>> > shell, and the other does all the work.)
>>
>> Hm... Sounds great :-) How do we achieve that? ;-)
>
> By the devices of a mighty magician called Dynamic Linking...

I see...

> Seriously, the details are still all very open. Moreover, they have been
> and are being discussed in other threads. No need to repeat it here.

Clear.

Regards,
scolobb