I highly recommend to stay compatible and add a utc timestampas a new field
and deprecate the old field for at least one release cyle.
On 15 Oct 2016 7:20 p.m., "JJoe2" <g...@git.apache.org> wrote:
Github user JJoe2 commented on the issue:
https://github.com/apache/log4net/pull/25
Thanks.
I agree that I should have done a better job of keeping these changes
separate but when I did the initial work I was very much a git novice.
I’ll bite the bullet and separate these changes into two branches with
a rebase on the latest trunk.
I take your point about the binary serialization format. For now I’ll
roll back the change in the serialization / deserialization methods. Users
of RemotingAppender will therefore still be subject to ambiguous timestamps
but it’s no longer a breaking change.
We could also consider implementing serialization as follows (perhaps
enabled by a configuration setting if preserving backwards compatibility is
important):
info.AddValue("TimeStamp", m_data.TimeStampUtc);
// Serialize
m_data.TimeStampUtc = info.GetDateTime("TimeStamp").ToUniversalTime();
// Deserialize
Given that DateTime.ToUniversalTime() is a noop if the Kind property is
already set to Utc, this would give the correct result if both sides have
the new version, and the following result (ignoring .NET 1.x which didn’t
have a Kind property) for mixed versions:
1. Old client and new server
Old client serializes as local time; New server converts this to UTC
based on the server’s timezone. Result unchanged as long as client and
server timezones are the same.
2. New client and old server
New client serializes as UTC. Old server will have a UTC time in the
TimeStamp, and will probably interpret it as local time.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
