1 |
On Wed, Nov 23, 2005 at 10:38:39AM -0500 or thereabouts, Daniel Ostrow wrote: |
2 |
> And herein I think lies some confusion. Personally if I were an AT both |
3 |
> would be important but more to the point the "more up to date" issue |
4 |
> would be the most important. |
5 |
|
6 |
I agree -- this was the main point of the original GLEP. |
7 |
|
8 |
> an AT does a |
9 |
> `cvs up` and retests to try and catch *other* errors all within a matter |
10 |
> of *single digit* minutes. |
11 |
|
12 |
I do question the need for "single digit" minutes. 30 minutes may be too |
13 |
much, but I think we could probably live with something in the 10-15 minute |
14 |
range. (if folks disagree, please speak up) |
15 |
|
16 |
> I know this is a far cry from what you are proposing, and I understand |
17 |
> that the present CVS server cannot handle this sort of load but I |
18 |
> believe that this was the original intention at least...someone correct |
19 |
> me if I am wrong. |
20 |
|
21 |
Anything is possible -- it's merely a matter of how much money we want to |
22 |
spend in the process. So far, nobody has really come back and said that |
23 |
using CVS, specifically, is a requirement. So, at this point, all options |
24 |
are on the table, but the main goal is to provide something that is as |
25 |
close to real-time as possible and allows authorized individuals to |
26 |
synchronize far more often than the current public rsync mirrors. All this |
27 |
is for a targeted group of up to ~100 users. |
28 |
|
29 |
Can we agree on these requirements? Are there others that I've left out? |
30 |
If not, we can start working on an implementation plan. |
31 |
|
32 |
--kurt |