[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Proposal to merge innosetup-hsuninsfix-5x-1.2.x without explicit +1

From: Kevin Puetz <puetzk_at_puetzk.org>
Date: 2005-06-13 07:40:25 CEST

Max Bowsher wrote:

> Currently we have a Windows installer bugfix in STATUS, lacking a +1, and
> unlikely to get one, since josander is taking a vacation from subversion,
> due to external time pressure, and we don't really have anyone else
> focusing on this area.
>
> However, josander noted in
> http://svn.haxx.se/dev/archive-2005-05/1117.shtml that the fix was used to
> produce the actual installer executables from 1.2.0-rc4, so under the
> circumstances, I am inclined to consider this change "approved by mass
> testing".
>
> Thoughts?
>
> Max.

Well, I'm neither a subversion committer nor particularly familiar with the
svn installer... but I do use InnoSetup at work. So a quick glance through
those changes:

The use of restartreplace on the uninstaller makes sense (else in modify
mode it will never get replaced; it's already open).

The use of {srcexe} likewise removes a dependency on the local filename for
setup.exe, which is good since users might rename it :-)

The [Components]/[InstallDelete] section is trying to work around the
behavior that 'disablenouninstallwarning' supressed the warning for.
Usually, when doing an upgrade or similar, not selecting a component that
was selected before won't remove those old files (nor would it upgrade
them, since that component was skipped this time around). This way they all
get removed at the beginning, and come back only if selected.

This solution isn't something I'd have chosen to do; keeping the list of all
files that any version ever shipped as part of that component usually gets
a bit unwieldy. But I don't have a better one (other than leaving the
warning on and saying "don't do that"), and it does work (assuming that the
list of filenames is maintained and is stable enough not to get too
unmanageable).

Finally, the combination of runminimized and runhidden seems odd to me (it
looks redundant), but it does match what's specified in the readme for
uninshs. I'm also not sure why one would use nowait; it would seem we
should wait for uninshs to finish whatever /r is asking it for before we go
on. So that part I can't really comment on...

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jun 13 07:41:40 2005

This is an archived mail posted to the Subversion Dev mailing list.