On 3/19/20, 12:17 AM, "Christofer Dutz" <[email protected]> wrote:
Good morning all :-)
Ok ... so I've just tested my changes in my fork and on my own maven
infrastructure. Worked like a charm.
Now after the "mvn release:prepare" I can see 2 committed, but not pushed
changes and I have to manually push them (including the tag)
After that the build now doesn't sign the artifacts.
But Apache's nexus won't let us release artifacts that are not signed. So
what is the process in signing them in the 13 steps?
The RM downloads, verifies (hopefully by building the source package locally
and doing a binary compare), then signs. There is an Ant script that automates
that.
I could simply not deploy to a remote maven repo but to a directory in the
local file-system and then all of them can be batch-signed (I hope)
Then you would use the wagon plugin to deploy the entire directory.
Was that the process?
Yes.
Chris
Am 19.03.20, 00:21 schrieb "Alex Harui" <[email protected]>:
Ideally, in the fixing and testing of the steps, a release would be
generated. The current steps will need adjusting since there is no utils
profile anymore.
-Alex
On 3/18/20, 3:00 PM, "Carlos Rovira" <[email protected]> wrote:
Hi Alex,
since I think we are doing circles, maybe the best way to solve, if
we have
into account that hopefully Chris can repair soon the CI steps, is
to ask
for someone that want to be then next RM using the current process
13 CI
step process to cut the 0.9.7 release.
What do you think?
Thanks
El mié., 18 mar. 2020 a las 21:44, Alex Harui
(<[email protected]>)
escribió:
> Because you keep proposing processes that don't actually use the
build.xml
> files to build the Ant artifacts. Your prior post is not a
correct summary
> of what I'm saying. The RM must run the build.xml files in
creating the
> artifacts to prove that they work.
>
> -Alex
>
> On 3/18/20, 1:40 PM, "Christofer Dutz"
<[email protected]> wrote:
>
> Gee ... I wanted to go to bed, but I had to respond to this.
>
> No one ever said to get rid of Ant ... why do you keep on
claiming
> this is the goal?
> We just want to create the release with only one of the two
and not
> with both.
>
> Chris
>
>
> Am 18.03.20, 21:37 schrieb "Alex Harui"
<[email protected]>:
>
> We already use Maven to release the Maven artifacts.
Maybe you
> can explain how the RM can test the validity of the Ant artifacts
without
> using Ant. Otherwise the RM has to run Ant at some point, and
that's why
> we currently use Ant in certain places and Maven in others. And
not get
> rid of Ant.
>
> -Alex
>
> On 3/18/20, 1:32 PM, "Josh Tynjala"
<[email protected]>
> wrote:
>
> I believe, as they do, that a Maven release process
will be
> easier.
>
> --
> Josh Tynjala
> Bowler Hat LLC <
>
https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbowlerhat.dev&data=02%7C01%7Caharui%40adobe.com%7Cec52b8f899ab4aba1e5f08d7cbd59cd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637201990638198549&sdata=kPQQgQ9nRbmoy4e%2BQe3kVDV4Enzot7tIQn1s2AN8SPk%3D&reserved=0
> >
>
>
> On Wed, Mar 18, 2020 at 1:31 PM Alex Harui
> <[email protected]> wrote:
>
> >
> >
> > On 3/18/20, 1:25 PM, "Josh Tynjala" <
> [email protected]> wrote:
> >
> > Make that two PMC members. If Carlos and Chris
can make
> it possible to
> > do
> > frequent releases, I want that too.
> >
> > We already had a way to do frequent release. So I
don't
> understand
> > spending more time to do something we already could
do.
> >
> > If what they come up can't produce a valid
release, then
> I agree that
> > that
> > would count as destroying existing work, and
that's not
> acceptable. But
> > that's not what they're planning to do.
> >
> > --
> > Josh Tynjala
> > Bowler Hat LLC <
> >
>
https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbowlerhat.dev&data=02%7C01%7Caharui%40adobe.com%7Cec52b8f899ab4aba1e5f08d7cbd59cd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637201990638208544&sdata=HaYOM3MRzwcsoMtqjabhqmj2N7dA1GeWUGuAgHE3IAY%3D&reserved=0
> > >
> >
> >
> > On Wed, Mar 18, 2020 at 1:14 PM Alex Harui
> <[email protected]>
> > wrote:
> >
> > >
> > >
> > > On 3/18/20, 12:06 PM, "Christofer Dutz" <
> [email protected]>
> > > wrote:
> > >
> > > Hi Alex,
> > >
> > > where is this thrown out?
> > >
> > >
> >
>
https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fapache%2Froyale-compiler%2Ftree%2Fdevelop%2Fcompiler-build-tools&data=02%7C01%7Caharui%40adobe.com%7Cec52b8f899ab4aba1e5f08d7cbd59cd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637201990638208544&sdata=jLBTFYM6Fds690vPJmRER1sg%2FW8YtfDZz%2F8ZpRVnH0k%3D&reserved=0
> > >
> > >
> >
>
https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fapache%2Froyale-compiler%2Ftree%2Fdevelop%2Fcompiler-jburg-types&data=02%7C01%7Caharui%40adobe.com%7Cec52b8f899ab4aba1e5f08d7cbd59cd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637201990638208544&sdata=KMyuRghmFZ0Ywd94CTjx7HoNjDLxP0mfWctWT5Hh%2By4%3D&reserved=0
> > >
> > > If you want to release them, all you need
to do is
> go into that
> > > directory and run:
> > >
> > > mvn release:prepare
> > > mvn release:perform
> > >
> > > So as you see nothing is removed ...
> > > there is no need to have them in the same
reactor,
> if that's
> > what you
> > > are referring to.
> > >
> > > The new documentation does not mention them,
so how
> would someone
> > know to
> > > do that? So sure, the new process seems
simple but
> that's because it
> > > doesn't document how to handle those two
modules.
> > >
> > > And the argument to not try something
again which
> had failed in
> > the
> > > past is sort of a strange argument for a
project
> discussing on going
> > 1.0
> > > I bet a lot of stuff is in Royale that
didn't work
> before and now
> > > works. It's not that we are doing it the same
way we
> did before.
> > > If we did, I agree that would be a stupid
waste of
> time.
> > >
> > > AFAICT, you are doing it the same way as
before, but I
> won't know
> > for sure
> > > until the set of steps to create a release are
> documented.
> > >
> > > But you didn't answer my question on
where it is
> written that
> > you must
> > > build jars used by Ant with Ant.
> > > Would you mind giving an answer on that?
> > >
> > > Ant should be used to test Ant. I don't
understand
> how you can test
> > Ant
> > > without running Ant.
> > >
> > > The Apache Way is about scratching your itch,
but it
> is also about
> > > consensus, and the itch you scratch should
not break
> or destroy
> > someone
> > > else's work. Otherwise you are saying that
someone
> can say "no,
> > Royale
> > > should not be PAYG" and rewrite the Basic
libraries
> and I can't do
> > anything
> > > about that. That's not how Apache projects
are
> supposed to work.
> > >
> > > We have one PMC member who wants to use Maven
to do
> things Ant
> > already
> > > does. Why bother? Ant is better for
scripting a set
> of steps.
> > That's why
> > > it exists.
> > >
> > > Let's fix the current steps and get a release
out. I
> think you'll
> > see
> > > that it uses Maven "appropriately". Then
make a
> branch and figure
> > out how
> > > you want to make a release. Because we have
> reproducible binaries,
> > you
> > > should be able to have your branch exactly
match then
> next release
> > so we
> > > know nothing got missed.
> > >
> > > My guess is that the net is going to be to
use Maven
> to tie a bunch
> > of
> > > steps together that are currently tied
together with
> Ant. Assuming
> > that
> > > Ant is still used to validate the build.xml
files and
> ant task jars.
> > >
> > > -Alex
> > >
> > >
> > >
> > >
> > >
> > >
> >
> >
> >
>
>
>
>
>
>
>
--
Carlos Rovira
https://nam04.safelinks.protection.outlook.com/?url=http%3A%2F%2Fabout.me%2Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com%7Cec52b8f899ab4aba1e5f08d7cbd59cd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637201990638208544&sdata=lVM%2FE1BSzPyPgVXRiv6dkFFCjl4u%2Bhwb7tr46UTZkL8%3D&reserved=0
