Re: A thought on Windows Experience

[Richard Shann]

On Mon, 2013-12-09 at 14:15 +0100, David Kastrup wrote:
> Richard Shann <address@hidden> writes:
> 
> > On Mon, 2013-12-09 at 11:51 +0100, David Kastrup wrote:
> >> David Kastrup <address@hidden> writes:
> >> 
> >> _Very_ frustrating and unusable.  Complains about missing libraries when
> >> starting but those are available in Ubuntu.

I missed this bit when I replied before. Remarkable that it does
actually start then, in fact, bizarre, this is not a static executable
but a dynamic one with a bunch of shared libraries bundled in (like
LilyPond, in fact, built with the same GUB machinery).


> >> 
> >> Opens what feels like dozens of overlapping windows you first need to
> >> cleanup.
> >
> > Yes, this has annoyed all ponders who have tried the latest Denemo. I
> > guess they will have to stay closed on the first run. (Once you close
> > them they stay closed if you quit cleanly). It will then be up to the
> > user to find the palettes in the View menu and start exploring them.
> >
> >>   Refuses to compile anything, stating in the print preview
> >
> > I wonder why - it works for some distros out of the box - otherwise you
> > have to give the path to LilyPond in Edit->Change Prefs->Externals
> 
> Nope.  Not even with an explicit path to either 2.19.0 or 2.16.2.  Just
> displays the following screen shot.
> 
> Also seems to have some memory managing problems: the fonts in the
> windows displayed "E" instead of "k".

This does indeed look like memory corruption - the other symptoms you
give are not like anything I have seen with Denemo running correctly
either.
I thought I would test out the binary that is on denemo.org and
downloaded it in a virtual machine running a vanilla Debian stable O/S,
the result: it will not even start. The executables ~/usr/bin/denemo and
~/usr/bin/lilypond are present and have the right permissions but
attempting to execute them from the bash prompt results in a baffling
"No such file or directory" message:

 ls -l denemo
-rwxr-xr-x 1 rshanngub rshanngub 1479840 Nov 25 23:00 denemo
address@hidden:~/denemo/usr/bin$ ./denemo
bash: ./denemo: No such file or directory
address@hidden:~/denemo/usr/bin$ 

and then the same thing for lilypond:

ls -l lilypond
-rwxr-xr-x 1 rshanngub rshanngub 4377128 Nov 25 23:00 lilypond
address@hidden:~/denemo/usr/bin$ ./lilypond
bash: ./lilypond: No such file or directory

This is something I've seen before and assumed was to do with my
development environment. I think it gets tested on Ubuntu distributions
before being uploaded, and certainly it seems to get a bit further on
your distro. I can only appeal to anyone with a GNU/Linux system to test
it out.

> 
> >> window that LilyPond can't compile stuff (there is a version of LilyPond
> >> in the path).  One can open the LilyPond source file window (looks like
> >> it should work in 2.19.0), but Denemo refuses to open the "LilyPond
> >> error" window.
> >
> > It does actually open the LilyPond Errors pane, but as Denemo is unable
> > to run LilyPond that is empty.
> > The LilyPond Errors item is not a separate window but a pane in the
> > LilyPond view, it would be better if the toggle for this lived in the
> > LilyPond window.
> 
> No, there is no LilyPond error view, not in a pane or otherwise.
> 
> > Well as this is a LilyPond output window with no LilyPond executable
> > found this is not by itself surprising. It should tell you (once only)
> > that it didn't find LilyPond.
> 
> It's a text in the window, and it does not change.
> 
> >> So much for the binary install.  I am not too enthused about the
> >> prospect of having to compile from source just to be able to test
> >> basic functionality and possibly get a better clue about the intended
> >> startup behavior.
> >
> > For folk with compilers, autotools and so on already installed
> > building from source is painless - it is not like running GUB. The
> > list of packages needed is on the Download page. (Hmm, pretty
> > painless, but there is some squabble amongst the distros about
> > splitting up one library into two ...)
> >
> > I have put in bug reports for the problems you have unearthed - Thank
> > you!
> 
> Do you have any users actually having success with the binary package on
> Ubuntu?  If not, telling people that the "ancient" versions delivered
> with the system itself are not to be used is creating a rather large
> barrier of entry.

As I say, I am told that it is a Ubuntu system that it gets tested on,
but if would help if we got more feedback from users. I occasionally get
visits from people with apple macs and the mac versions have worked on
their machines and I test windows versions on two or three machines with
various flavors of their o/ses. A GNU/Linux binary you would have
thought would be easier than either of those to get working ...
(As I wrote this I recalled that I have a small partition with Ubuntu
12.04 installed, I rebooted and went through the same process as with
Debian Stable and got the same, bizarre result). I think we need to warn
people that they may well need to built it :(

Richard