In D3756#58937, @indygreg wrote:

I'm unsure about this change. On one hand, the comment (which appears to have been added to mpm several years ago) implies that we never should have generated backups in this case. On the other, one of the key rules of a VCS is "don't eat my data." Even though we are aborting the operation, I could see some scenarios where someone would want a backup of the aborted/partially-completed rebase.

@durin42: what do you think?

FYI, this is intended for the stable branch.

@indygreg I'm not sure if this undermines your testing strategy at all, but if it does we can try and figure out another path forward...

In D3694#58369, @lothiraldan wrote:

How much work is this, do you have any idea?

The first option (adding "reserved" phase) should be very quick to implement. It might need minor adjustment for performance but I don't expect many.

The second option (changing all algorithm to handle the gap) is more work since about all algorithm touching phases in Core and extensions assume they can be handled as a simple list.

So I would pick the first option.

Good suggestions, integrated them. :)

In D3700#58403, @lothiraldan wrote:

In D3700#58376, @durin42 wrote:

I see some what, but not any why. Why is this useful?

I need this changeset to integrate the mercurial test runner with some external tools.

I'm in favor, but feel like I've got enough conflict of interest I shouldn't land the patches.

I see some what, but not any why. Why is this useful?

In D3694#58241, @lothiraldan wrote:

Interesting. I think I like this, it's a bummer that it requires a format bump in requires.

• There are no format changes per-se, older client would preserve the phases for internal changesets, but have them visible,

In D3649#57906, @yuja wrote:

(resend because I forgot to remove the Phab shit, sorry for the noise.)

> It's surprising that `hg serve --print-url` spawns a daemon process!

If it does not launch a daemon process, the procress caller (`xdg-open`, `midori`, `firefox`, etc.) waits the end of the stream to launch, so it waits that the `hg serve` procress ends!

You can just run hg serve -d --print-url.

The option could be named as --daemon-print-url if we needed such big
red switch, but I don't think that makes sense.

Lovely. Queued for stable, many thanks.

In D3684#58074, @lothiraldan wrote:

We have an implementation of shelve without strip nor obsmarkers.