In Mon, 8 Apr 2019 18:59:21 +0100
Will Godfrey <willgodfrey@xxxxxxxxxxxxxxx> wrote:
On Sat, 23 Jun 2018 18:40:51 +0100
Will Godfrey <willgodfrey@xxxxxxxxxxxxxxx> wrote:
On Sat, 23 Jun 2018 17:16:30 +0500
"Nikita Zlobin" <dmarc-noreply@xxxxxxxxxxxxx> (Redacted sender
"cook60020tmp" for DMARC) wrote:
i looked some xiz files from latest zyn - and it seems to be
compressed xml. Is not it possible yoshimi-spec data in special
block, like extension, which would not break zyn?
This shouldn't be a problem. Last time I checked, both Yoshi and Zyn
simply ignored entries in the files that they didn't recognise. When
Yoshi saves a .xiz file it is completely Zyn compatible. Also, Zyn
simply won't see .xiy files at all.
Wait... what if xiz file is saved from zyn, and than you need to
load to yoshimi again? It could just check, what version is newer,
and update older - may be ask about user choice in this case. As
fariant, load yoshimi version anyway, but if xiz appeared newer,
than ask, wether to update it (only after loading).
Now that is something I didn't think of :(
However, date/time stamps might not be very reliable as an
indicator. Possibly better would be reading the info tags. Ones
saved by Zyn won't have the Yoshi version numbers. It does rather
slow things down though.
It's a completely valid point, but I don't know how far we should go
in protecting users.
Will.
Coming back to this after a loooong time. I think I have an answer.
First of all, always save both formats - these days disk space is not
exactly tight.
Keep in mind that an instrument can't be shown in a bank if it doesn't
actually exist in either form.
When loading an instrument, first look to see if there is an 'xiy'
file. If not simply load the 'xiz' one.
If there is an 'xiy' one, then look for the 'xiz' one. Again, if it
doesn't exist simply load the 'xiy' one.
If there is an 'xiz' one, look at the header. If there is a
'Yoshimi-author' field then it must have been saved by Yoshi so load
the 'xiy' one.
If that field is missing, it's been saved by Zyn, so at that point
notify the user and offer a choice.
The only time this fails, is if a program change is called via MIDI,
in which case I think we simply prefer the 'xiy' one.