1 |
On Thu, 13 Aug 2015 11:36:58 -0700 |
2 |
Zac Medico <zmedico@g.o> wrote: |
3 |
|
4 |
> On 08/13/2015 02:06 AM, Alexander Berntsen wrote: |
5 |
> > Superficially it looks OK, except a few minor details. |
6 |
> > |
7 |
> > On 13/08/15 10:34, Zac Medico wrote: |
8 |
> >> Repos are synced in parallel (includes their post-sync hooks). |
9 |
> >> Output of concurrent processes is currently mixed (irrelevant with |
10 |
> >> --quiet). Support for FEATURES=metadata-transfer is handled in the |
11 |
> >> main process, which may be required for some backends (such as |
12 |
> >> sqlite). Repos are synced only after their master(s) have synced |
13 |
> >> (in case that matters for hooks). |
14 |
> > The mix of tenses in here is making it difficult to read for me. The |
15 |
> > present tense arguably should not be used since it is ambiguous. |
16 |
> > Does it mean pre-patch or post-patch? Difficult to say. Please use |
17 |
> > the imperative mood to dictate what a patch does. |
18 |
> |
19 |
> How's this future tense version? |
20 |
> |
21 |
> Repos will now be synced in parallel (including their post-sync |
22 |
> hooks), but a given repo will only be synced after its master(s) have |
23 |
> synced (in case that matters for hooks). Output of concurrent |
24 |
> processes will be mixed (irrelevant with --quiet). Support for |
25 |
> FEATURES=metadata-transfer will be handled in the main process, which |
26 |
> may be required for some backends (such as sqlite). |
27 |
> |
28 |
> |
29 |
> |
30 |
> >> +class SyncScheduler(AsyncScheduler): |
31 |
> > + ''' |
32 |
> > + Sync repos in parallel, but don't sync a given repo until |
33 |
> > all |
34 |
> > + of it's masters have synced. |
35 |
> > s/it's/its/ |
36 |
> |
37 |
> Thanks, fixed. |
38 |
> |
39 |
> > |
40 |
> > I tried applying it and testing it, and it seems to work for me. |
41 |
> > |
42 |
> |
43 |
> Excellent! |
44 |
|
45 |
|
46 |
That looks good to me :) |
47 |
|
48 |
merge away :) |
49 |
|
50 |
-- |
51 |
Brian Dolbec <dolsen> |