On 18/11/2019 22:01, Rémy Maucherat wrote: > On Mon, Nov 18, 2019 at 10:22 PM Mark Thomas <ma...@apache.org > <mailto:ma...@apache.org>> wrote: > > On 17/11/2019 19:01, Mark Thomas wrote: > > We have a regression affecting 8.5.x (and 7.0.x). > > The fix [1] for an issue raised on the users list [2] was incorrect and > creating a different issue [3]. > > There are two questions to address. > > a) Do we cancel the 8.5.49 release for this and roll a 8.5.50 release? > > b) Do we fix this by a) correcting [1] or back-porting the change to a > single poll set? > > I'm leaning towards 8.5.50 and back-port the single poll set change. > > Thoughts? > > Mark > > [1] > > https://github.com/apache/tomcat/commit/fffb08790e642e03f00c5f96a3a61ee09a2c8342 > [2] https://markmail.org/thread/lh7pst3bwptpcyco > [3] https://tomcat.markmail.org/thread/btjionfhp2olyjne > > > +0 only for 8.5.50 since the issue isn't so common (APR + Windows).
Ack. > Is porting the multipoller removal to 7.0 really doable ? I don't know. I haven't looked at the code yet. I do have an alternative fix I tested that should work if back-porting proves tricky. > It could be running on older Windows platforms maybe. That would only be an issue of the versions of Windows were so old they were no longer supported by Microsoft. Users in those circumstances are unlikely to be updating to the latest 7.0.x release anyway. Therefore that isn't something I'm particularly worrying about. (But I am happy to be over-ruled if the consensus is not to back-port to 7.0.x.) Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
