| 1 |
On Sun, 30 Dec 2007 20:11:16 -0800 |
| 2 |
"Alec Warner" <antarus@g.o> wrote: |
| 3 |
> On 12/30/07, Ciaran McCreesh <ciaran.mccreesh@×××××××××××××.uk> wrote: |
| 4 |
> > Is it legal for ebuilds to call has_version and friends in |
| 5 |
> > parallel? Is it legal for ebuilds to call has_version and friends |
| 6 |
> > after the ebuild process has terminated? Discuss. |
| 7 |
> |
| 8 |
> If the pm implements read/write locking on the underlying datastore |
| 9 |
> (which it should probably have regardless of this request) then I |
| 10 |
> don't see a problem in parallel has_version calls. |
| 11 |
|
| 12 |
Actually, it's the communication channel that's the issue... If, for |
| 13 |
example, has_version is implemented in terms of a request on a pipe |
| 14 |
rather than execing a new package manager, we get into messy bash |
| 15 |
locking territory... |
| 16 |
|
| 17 |
> I don't get your second example..do you mean the ebuild is running |
| 18 |
> has_version in the background and then terminating? |
| 19 |
|
| 20 |
Yeah. Again, consider the pipe example. If the package manager closes |
| 21 |
off the pipe when it thinks the ebuild's done, calling has_version will |
| 22 |
get the backgrounded process SIGPIPEd. |
| 23 |
|
| 24 |
-- |
| 25 |
Ciaran McCreesh |
Archives