Am 07.04.2018 um 00:48 schrieb Richard Kimberly Heck:
installer that is as automatic as it can be. Many of the rest of us have
real doubts about the wisdom of this,
That is what annoys me. You doubt that I am wise enough to set up LyX
for Windows users.
First of all, I specifically said that THIS IS NOT THE ISSUE WE ARE
DISCUSSING.
This IS the issue for me because you state my solution does harm users
and it is not wise. Fine, then dive into the Windows ocean and propose a
new/better Win installer that acts wisely.
What I doubt is that it is a good idea for one
person (you) to try all by yourself to manage a large collection of
complex programs and their interactions and dependencies.
The installer code in in our Git. Everybody can contribute to it and I
hoped that I won't be the only contributor.
That the installer works as it does is the result of about 10 years of
development, many, many discussions. So please don't come up now with
doubts that its concept is not wise. Se also my other mail from today.
That leads inevitably to the kinds of
problems we have here,
to multiple versions of the installer in response to bugs in programs we
do not control, and so on and so forth.
LyX needs many third party programs. For the user it is irrelevant if he
cannot use LyX because of a problem with python, Ghostscript, Perl,
MiKTeX, NSIS or ImageMagick. He only sees the result: LyX is not working
well for me. Therefore the lyX Win installer must take care of the
dependencies. This is the conclusion of 10 years LyX Win installer.
So let's face the reality and act accordingly.
As JMarc pointed out, the very example you give:
It is wrong to say, LyX is not allowed to update MiKTeX because of
potential bugs on CTAN. I mean with this directive on Linux they cannot upgrade
e.g. my
Qt5 from e.g. 5.6 to 5.9 because there could be a regression bug in Qt5.9 that
would
affect all apps using Qt5.
just makes this point. People can do these kinds of upgrades if they
like, but a stable distro
like Debian or Ubuntu LTS...
Again, we must face the reality! Win 10 is a service. If you like it or
not, you will be upgraded to current Win 10 builds. The same is with
MiKTeX. As soon as you upgrade you need the new package handling system
to be able to get bugfixes in future. Moreover, MiKTeX cannot stand
still because some new Win 10 features/settings require a changed
MiKTeX. Imagine people got their Win 10 upgrade and cannot use MiKTeX
any longer. (The MiKTeX developer is for good reason in the Windows
early tester program.) Therefore I don't understand the whole discussion
why an upgrade to the new MiKTeX package handling could be harmful.
So your Linux-centric view doesn't help us for LyX on Windows. Windows
is the market leader and we have to accept that the Windows world works
in some respects even fundamentally different than the Linux world.
(Besides this, popular Linux like e.g. Manjaro upgrade permanently e.g.
Qt. If I like it or not, I cannot prevent this.)
So my final question still stands:
_Should I remove the Win installer code from LyX's git?_
If yes, I'll release installer builds as I di the last years.
If no, I'll create a new Github project and LyX won't have an "official"
Win installer but refer to external LyX win installer.
Please make your decision and tell me.
many thanks and regards
Uwe
