Btw. just so we compare the right apples and oranges -
which target file do you want me to use ? also .multiple ?
Yes, multiple only. The one that is recommended to use for
development in IDE as it provides source "naturally", and which is
used by Tycho to create the mirror.
Since I can also use the mvn output as a mirror setup
it at least for me is much faster and reputable.
Why do you need a mirror? Tycho caches so much stuff that you
already have a mirror without even configuring one.
btw. if tycho is so much slower - I wonder what it is doing since
PDE should be doing the same thing...
PDE relies on p2, which relies on ECF, allowing parallel downloads
among other optimizations. I guess It also fetches only p2 artifacts
and unpack them locally, dividing by 3 to 4 the amount of downloaded
bytes. It's just smarter than the mirror goal (that is not part of
Tycho, but a plugin of ours).
You want me to report broken network connections to
eclipse.org mirrored servers ?
So broken connection are not at all related to how you consume TP.
They happen for both "mvn verify" and IDE the same way. This is not
a criterion to tell one thing is better than the other.
FYI, I also often have broken connection when mirroring TP with
Maven. It's not better for that.
If you feel comfortable with TP, and want to hack stuff, it's up to
you. But beware that several people in the team didn't fully
understand it yet and get confused by opposite instructions. The
simplest approach is the one documented in
https://github.com/jbosstools/jbosstools-devdoc/blob/master/building/target_platforms/target_platforms_for_consumers.md#load-in-ide
, which as for only drawback compared to mirroring locally to make
workspace not usable for the time it fetches artifacts. But it's
simple, people understand it, and it only cost 15 minutes per
workspace every 6 weeks or so.
I believe there aren't many arguments remaining for not recommending
simply using .target file in IDE :P