On Saturday 21 January 2006 17:37, Chris Cannam wrote:

> I suppose what makes me glum is the thought that there is no fix for
> some of these problems (other than switching to a whole different
> distro), and that other problems sound really simple but will be
> confusing for the user to fix even if we do identify them.

True, but we can't cure all of that from within Rosegarden.  We should do what 
we can.  I know it all feels like a gigantic pain in the ass (arse), but 
answering the same questions ad nauseum is a gigantic pain in the ass (arse) 
too.  I've lost any hope of any of this magically becoming easier to deal 
with in some reasonable timeframe.

> Still, let's keep on the bright side.  Anything is better than nothing.
> Another point: if we can establish that everything _is_ in fact OK,
> it's got to be worth telling the user that.  That's especially useful

Yes, I like that idea.  "Congratulations!  You have a Rosegarden Ready(tm) 
system!  Here's your certificate of readiness, and a coupon worth one beer at 
the regular price at your local watering hole, unless your name is Guillaume 
Laurent, in which case this certificate is valid only for orange juice."

> if they've been working to fix problems they had previously.
> (Question: should we tell them about it only the first time they run
> RG, and when they select a "Configuration Wizard" or some such option
> subsequently, or should we tell them every time they start up RG, but
> with one of those [X] Don't tell me this again options?)

I like the K3B way.  It looks every time it runs, and it doesn't say anything 
if there is no problem, but then it goes back into "I found trouble" mode as 
soon as a problem comes back up (which often happens after I upgrade, as the 
default installed bits have permissions that make K3B unhappy.)  This model 
might be a little too simplistic for what we're dealing with here, since K3B 
is primarily just checking permissions on a few files, and ensuring that this 
and that helper apps are installed properly.

I think I'd want some detection/flag logic so that any "you're certified" 
dialog only comes up a) the first time you ever run, or b) after it is 
detected that you had a problem, and it's now cured.  I think the "you have a 
problem" warnings should all be don't-show-able, so someone can say "OK, I 
don't care, I don't want to run JACK anyway, and I don't mind if my timing 
resolution sucks" or what have you.

As far as this first-time thing, some other things to do here might be "Here's 
a sample of thus and such (print preview?). Do the fonts look OK?  If not, 
here's our hack workaround font dialog.  Go play with it and find something 
that works."  (That problem G hacked around just before 1.0 by inserting 
user-configurable fonts for some reason or other.  The details escape me, but 
I *think* I still have non-default fonts configured for something as a 
consequence.)  Something like this, if done at all, should really be a 
*first* time, and not a recurring analyze the system and report problems kind 
of thing.

Also I guess this will undoubtedly slow down startup, so there should be an 
option to turn off the whole startup check process.  Maybe making the wizard 
something they could run on demand.  Or maybe it should always run on demand 
only.

I'm just thinking and typing in a straight line here, so forgive me if I'm 
rambling.  I'm also drinking Guillaume's share of beer.

> > kernel they're running (this one has been cropping up quite a lot
> > since people started switching to udev).
>
> Oh, is *that* why this happens?

Seems to be, yes, although that's merely something I've gleaned.  Seems to be 
an issue with new installs that prefer udev, and udev not doing the right 
thing as far as providing MIDI infrastructure.  I have the vague, half-formed 
impression that it's probably going to come up as often as the kernel timer 
thing.

> > We can detect if they have an emu10k1, [...]
> > We can detect if they have an AC97 or other card with a hardware
> > port, and just simply *ask* them if they have a MIDI keyboard or

> Yes, fair enough.

Although all that about the emu10k1 I was throwing out yesterday may well just 
be pretty worthless after further consideration.  Not many have them now, and 
nobody will have them in the future.  It's probably not a case worth wasting 
a lot of effort on, really.

> And kernels from 2.6.13 onwards won't get decent MIDI timing out of
> Rosegarden at all.  So that eliminates... all of them!  Let's just
> sling up a "Your kernel is not appropriate for audio use" message at
> startup whatever the hell they're using.

That would get it right most of the time, I'm afraid.  :(

> > 1) Spelling out "bliff" is tricky, admittedly
>
> B-L-I-F-F.

*chortle*

-- 
D. Michael 'Silvan' McIntyre  ----   Silvan <[EMAIL PROTECTED]>
Linux fanatic, and certified Geek;  registered Linux user #243621


Author of Rosegarden Companion http://rosegarden.sourceforge.net/tutorial/


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
_______________________________________________
Rosegarden-devel mailing list
[email protected] - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

Reply via email to