On 2016-12-14 01:13 , René J.V. Bertin wrote:
On Wednesday December 14 2016 00:26:18 Joshua Root wrote:
It was probably always fairly safe to interrupt rev-upgrade in report
mode. Any database writes it does to update the binary flag are in a
sqlite transaction.
I bet, but I've already found myself with a corrupted registry which can only
have happened during an sqlite transaction. I presume...
I often miss an option to skip that step (esp. the first of the day which
always takes longer) but I suppose that being able to interrupt it safely will
work just as fine.
The install and upgrade actions take a --no-rev-upgrade option.
Ah, I wasn't aware of that. But why the install action too? I would have
expected it do run the step (and did expect it until recently) but I've never
seen it do an actual rev-upgrade run. At least that's consistent with the
feature's name ;)
I couldn't quite parse that sentence, but install does indeed do
rev-upgrade at the end.
Is there an abbreviated version?
Every long option can be abbreviated as much as is unambiguous.
Macports.conf has a revupgrade_autorun option. But if you mean "miss" as
in "forget to use" then I guess no amount of options will help :).
No, though you're right that maybe what I actually would like is a way to run
the step on demand only (--rev-upgrade).
Forgetting is exactly the reason why I haven't yet disabled the feature in
macports.conf . Most of the time I don't mind too much to wait for a run, and
often I appreciate it, so the balance is still in favour of keeping it on.
You can run it on demand with 'port rev-upgrade'.
- Josh
