Okay. Thanks for the feedback - today, I learned a lot about the insides of Javamail :)
I have updated my PR:
- Updated README.txt to contain some documentation about setting a
custom ssl socket factory
- Dropped TLSv1 in the fallback protocols (if no custom set properties
are present)
Thanks,
Richard
Am Mittwoch, den 02.12.2020, 15:29 +0100 schrieb Romain Manni-Bucau:
> Guess you can just create a readme in the geronimo-javamail root
> project, will be sufficient as a first step.
> Abou he default I wonder if dropping tlsv1 cant be good since it will
> be dropped soon?
> Otherwise just adding the missing "o" in protocols i'm fine with your
> proposal.
>
> We need to refine if we do a javamail subsite or a generic spec
> subsite sill :s.
>
> Romain Manni-Bucau
> @rmannibucau | Blog | Old Blog | Github | LinkedIn | Book
>
>
> Le mer. 2 déc. 2020 à 15:26, Zowalla, Richard <
> richard.zowa...@hs-heilbronn.de> a écrit :
> > I updated the diff (cf. v2) to (hopefully) address the concerns
> > raised
> > (if I understood them correctly).
> >
> > If you point me to a location where I can add a README /
> > documentation,
> > I would be happy to fill another JIRA with a related PR to document
> > the
> > usage of the custom ssl socket factory.
> >
> > Am Mittwoch, den 02.12.2020, 13:58 +0000 schrieb Zowalla, Richard:
> > > Thanks for your thoughs - I think, I get the idea.
> > >
> > > Maybe:
> > >
> > > - Using "mail.smtp.ssl.protocls" to allow easier configuration
> > (as
> > > proposed in the PR) for MailConnection#getConnectedTLSSocket() -
> > > would
> > > address 1.
> > >
> > > - To address 3. and pre-claim: PR would enable all protocols;
> > maybe
> > > address this concern by adding a default fallback pointing to
> > TLSv1,
> > > TLSv1.1, TLSv1.2 and TLS v1.3 (if supported) if no custom
> > > configuration
> > > via "mail.smtp.ssl.protocls" is present?
> > >
> > > - Documentation is always appreciated ;)
> > >
> > > Wdyt?
> > >
> > > Am Mittwoch, den 02.12.2020, 14:41 +0100 schrieb Romain Manni-
> > Bucau:
> > > > Yes but issue that we don't want to enable them all too.
> > > > So to be concrete what about:
> > > >
> > > > 1. Enable a smoother configuration (to avoid a custom class)
> > > > 2. Document the custom class case better (at least in a readme)
> > > > 3. Change a bit default to inherit JVM ones
> > > >
> > > > Think we should make the 3 to consider this case treated (does
> > not
> > > > mean it must be in the same PR but more before next release).
> > > > Wdyt?
> > > >
> > > > Romain Manni-Bucau
> > > > @rmannibucau | Blog | Old Blog | Github | LinkedIn | Book
> > > >
> > > >
> > > > Le mer. 2 déc. 2020 à 13:20, Zowalla, Richard <
> > > > richard.zowa...@hs-heilbronn.de> a écrit :
> > > > > Ah sorry - I misunderstood your comment.
> > > > >
> > > > > A custom socket factory would indeed fix the problem, but it
> > is
> > > > > rather undocumented.
> > > > >
> > > > > Nevertheless I think, that the default fallback shouldn't be
> > > > > hardcoded or at least support some more protocols...
> > > > >
> > > > > Best and thanks for the idea,
> > > > > Richard
> > > > >
> > > > > Am Mittwoch, den 02.12.2020, 12:16 +0000 schrieb Zowalla,
> > > > > Richard:
> > > > > > Honestly I didn't. I discovered the hard-coded
> > > > > > String[]("TLSv1")
> > > > > > in
> > > > > > MailConnection#getConnectedTLSSocket(), which is (imho) a
> > bit
> > > > > > odd.
> > > > > >
> > > > > > Imho, users should either be allowed to specify the enabled
> > > > > > (and
> > > > > > supported) protocols or to use the default ones provided by
> > the
> > > > > > jdk
> > > > > > classes :)
> > > > > >
> > > > > > This is already done for
> > MailConnection#getConnectedSSLSocket
> > > > > > but
> > > > > > not
> > > > > > for the TLS handling.
> > > > > >
> > > > > >
> > > > > > Am Mittwoch, den 02.12.2020, 13:09 +0100 schrieb Romain
> > Manni-
> > > > > > Bucau:
> > > > > > > Hi Richard,
> > > > > > >
> > > > > > > Did you try a custom socket factory? In such a case you
> > fully
> > > > > > > control
> > > > > > > it.
> > > > > > >
> > > > > > > Romain Manni-Bucau
> > > > > > > @rmannibucau | Blog | Old Blog | Github | LinkedIn |
> > Book
> > > > > > >
> > > > > > >
> > > > > > > Le mer. 2 déc. 2020 à 13:01, Zowalla, Richard <
> > > > > > > richard.zowa...@hs-heilbronn.de
> > > > > > > > a écrit :
> > > > > > > > Hi,
> > > > > > > >
> > > > > > > > I did some debugging and found, that TLSv1 is hard-
> > coded in
> > > > > > > > MailConnection.java in v1.0.0 of Geronimo Java Mail.
> > > > > > > >
> > > > > > > > I filled a JIRA [1], which contains a patch proposal.
> > > > > > > >
> > > > > > > > Happy to receive some feedback.
> > > > > > > >
> > > > > > > > Thanks in advance,
> > > > > > > > Richard
> > > > > > > >
> > > > > > > > [1]
> > > > > > > > https://issues.apache.org/jira/browse/GERONIMO-6792
> > > > > > > >
> > > > > > > >
> > > > >
> > > > > --
> > > > >
> > > > > Richard Zowalla, M.Sc.
> > > > > Research Associate, PhD Student | Medical Informatics
> > > > >
> > > > > Hochschule Heilbronn – University of Applied Sciences
> > > > > Max-Planck-Str. 39
> > > > > D-74081 Heilbronn
> > > > > phone: +49 7131 504 6791
> > > > > mail: richard.zowa...@hs-heilbronn.de
> > > > > web: https://www.mi.hs-heilbronn.de/
--
Richard Zowalla, M.Sc.Research Associate, PhD Student | Medical
Informatics
Hochschule Heilbronn – University of Applied SciencesMax-Planck-Str.
39 D-74081 Heilbronn phone: +49 7131 504 6791mail: richard.zowalla@hs-
heilbronn.deweb: https://www.mi.hs-heilbronn.de/
smime.p7s
Description: S/MIME cryptographic signature
