D4860: repository: define and use revision flag constants

[martinvonz (Martin von Zweigbergk)]

martinvonz added inline comments.

INLINE COMMENTS

> indygreg wrote in changegroups.txt:139-142
> You mean removing the single integer flags field from a future changegroup v4? TBH I'm not sure we'd want to do that: bitwise fields do work for what we're using them for. It's just that we need better definitions around what flags are in play. e.g. the server should advertise revision flags that may be sent and clients should check for compatibility *before* attempting to pull data. And we need to abstract the flags in code such that we're not passing wire protocol bit flags down to the storage layer for verbatim storage, as that constricts behavior of storage backends.

I just mean that the "externally stored" flag doesn't seem to make sense in the wire protocol, but maybe I'm wrong?

PS. I wasn't blocking this patch based on that comment. I just happened to run out of time while reviewing this patch (but finished reviewing the earlier patches and queued them). I'll review this patch once I'm done with Boris' introrev() patches.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D4860

To: indygreg, #hg-reviewers
Cc: martinvonz, mercurial-devel