| 1 |
On Wednesday 25 June 2003 00.33, Anthony Floyd wrote: |
| 2 |
> [snip] |
| 3 |
> |
| 4 |
> >tony@power tony $ time emerge -S zope |
| 5 |
> |
| 6 |
> [snip] |
| 7 |
> |
| 8 |
> >real 1m23.409s |
| 9 |
> >user 0m26.580s |
| 10 |
> >sys 0m11.130s |
| 11 |
> |
| 12 |
> How important is this? Well consider the following on an "old" |
| 13 |
> machine (PP200): |
| 14 |
> |
| 15 |
> # time emerge -S zope |
| 16 |
> [snip] |
| 17 |
> real 6m38.645s |
| 18 |
> user 2m16.580s |
| 19 |
> sys 0m10.930s |
| 20 |
> |
| 21 |
> So, while I agree that this pushes the bounds of patience, I don't want |
| 22 |
> to be required to install MySQL or any other db engine just to run |
| 23 |
> portage. I'm sure many others don't want that other. However, having |
| 24 |
> an optional component, say capable of handling multiple databases and |
| 25 |
> activated with USE flags ... that makes sense. Ideally, optimizing the |
| 26 |
> current system without requiring a massive db engine would be better, |
| 27 |
> though. |
| 28 |
The problem is that the current system doesn't scale well as it is all text |
| 29 |
based files. There is already caching so if you make the same query and it |
| 30 |
is still in the cache things go a lot faster. It is also creating a lot of |
| 31 |
unnecessary network traffic where portage isn't local to the machine. Not a |
| 32 |
big issue yet most likely but it could be an issue when more media streaming |
| 33 |
comes into play. |
| 34 |
|
| 35 |
tony |
| 36 |
-- |
| 37 |
Contract ASIC and FPGA design. |
| 38 |
Telephone +46 702 894 667 |
| 39 |
http://pgp.mit.edu:11371/pks/lookup?op=vindex&search=0x633E2623 |
| 40 |
|
| 41 |
|
| 42 |
|
| 43 |
-- |
| 44 |
gentoo-dev@g.o mailing list |
Archives