| 1 |
On 5/26/20 12:48 AM, Michał Górny wrote: |
| 2 |
> On Mon, 2020-05-25 at 21:31 -0700, Zac Medico wrote: |
| 3 |
>> Since variables like A and AA can contain extremely large values which |
| 4 |
>> may trigger E2BIG errors during attempts to execute subprocesses, delay |
| 5 |
>> export until the last moment, and unexport when appropriate. |
| 6 |
>> |
| 7 |
> |
| 8 |
> Please don't. This will only hide the problem from developers who will |
| 9 |
> unknowingly commit broken ebuilds and cause users of alternative package |
| 10 |
> managers to suffer. |
| 11 |
|
| 12 |
We've seen in https://bugs.gentoo.org/719202 that developers can already |
| 13 |
do that with existing versions of portage, since the failure can be |
| 14 |
dependent on USE configuration. |
| 15 |
|
| 16 |
If we want to proactively prevent this issue, then we need to set a hard |
| 17 |
limit on the size of A and abort before we generate the manifest. |
| 18 |
-- |
| 19 |
Thanks, |
| 20 |
Zac |
Archives