| 1 |
On Monday 04 October 2004 3:58 pm, Nicholas Jones wrote: |
| 2 |
> > What's this I've heard about Portage's performance not being a |
| 3 |
> > real issue? (argument for not rewriting it in C/etc) |
| 4 |
> |
| 5 |
> Short version: C is my prefered language, what I code in at work, |
| 6 |
> and I think it's completely wrong for portage and wide deployment |
| 7 |
> of an actively changing system such as portage. It's more important |
| 8 |
> for portage to work than to be blazingly fast. |
| 9 |
> |
| 10 |
|
| 11 |
I wasn't arguing that Portage should be in C. I was just pointing out that |
| 12 |
since there's no real performance gains from such a conversion, it's not |
| 13 |
likely that changing the directory structure would hurt performance either. |
| 14 |
|
| 15 |
> > should finally implement fetch-on-demand... |
| 16 |
> |
| 17 |
> So instead of a sync connection and update, we can trade off a |
| 18 |
> complete tree for 1-to-N connections, comparisons, and fetches. |
| 19 |
> |
| 20 |
> I'd recommend passing this through infra to make sure that |
| 21 |
> causing an extra N-times the connections is reasonable. |
| 22 |
> |
| 23 |
> Also, you potentially cause the mirror to reject you as there |
| 24 |
> could be a great deal more competition on any given mirror. |
| 25 |
> Portage would be forced to add logic to repeatedly poll for a |
| 26 |
> free connection once it has started. If you get stuck behind |
| 27 |
> Dialup users, you could very well be there a while. Basically, |
| 28 |
> it can cause a race condition against servers. |
| 29 |
|
| 30 |
It seems like you're assuming that only one file can be fetched per |
| 31 |
connection. I haven't researched the topic much, but I'm pretty sure HTTP, |
| 32 |
not to mention FTP, supports multiple requests with one connection. |
| 33 |
-- |
| 34 |
Luke-Jr |
| 35 |
Developer, Utopios |
| 36 |
http://utopios.org/ |
Archives