Re: Hurdish applications for persistence

[Jonathan S. Shapiro]

On Tue, 2005-10-11 at 15:56 +0200, address@hidden wrote:
>    A problem here is that programs aren't and shouldn't be written
>    solely for the Hurd.
> 
> They should! What would the point be of the GNU system if they aren't?
> This is a GNU project after all, and the goal is to produce a concise
> system, the GNU system.  Not a bunch of programs that can run on any
> platform out there.  The only thing one should worry about is
> portability across architectures, but across systems it is pointless.

As a practical matter, I think that the truth will be a compromise. The
test will be:

  For the applications that we want to port, can we port them with
  acceptable engineering cost.

For example, I would be willing to remove gets() from the C library,
even though that is incompatible.

The EROS project has insufficient experience with these issues. Marc
Seaborn made a significant advance with PLASH, because he demonstrated
that the file system interface isn't completely fatal (which we had
previously believed).


shap