live.thisdevice + pattrstorage crash problem

[pid]

hi.

i have a situation with loading prefs to pattrstorage with the defered bang of a live.thisdevice. it always crashes live but not the max editor. had to load device with command-shift trick to debug.

here is reproducable crash situation:


<pre><code>
----------begin_max5_patcher----------
440.3ocqTssahCCD8YmuBK+LKJNPWxtuseGUUHmjYAWkXirGnPq5+9N9RZgV
PBP6CIVyYNY7YNdbdqfIZr6Auf+a9ibF6sBFKBE.X4XlXPsusW4izDF3Eayy
hIoTHrGiv85cvTbs12A6zsvX9+ZMnW+JD3HqlVlg2nv10ZypkNnES6dU4LJM
sTFVpeHELsj+T9aLaGzld.ixPlA0cwcmTzOjGumF0PbOE+woU8hOKgcKNVip
LZBBOrARJQznLqDS3Bsgz1SARuWTDdM4J8nAv6UqfuYRpVjrI9vAUiGcgHqg
S8YB+EmFgAaGvk2u88fLtLKFIquj+UcF+q9lcP4kbv6y1NczZmx8gFFNPsL5
HS6LtZNk0QVN+bLuYWbVUb7SN+1mBW7q+mlXNkWsC5VRkmD5xP6oa1hoqsrO
7RlnsWCF7n1Y9D9b5YwhJ50OqGqGUvja8cpxxp5uvcixondVigQy.4xSx.H3
VBFUSeJYLW7fe73OVJ5GDlu9il3AP.+zYBucqqczDx2r4edBzAdTaTg6NGyo
9TVq0ccf4H4JFzcarzU5rHtvD5Uqo5qQUgog6VTTv6E+CXIrawD
-----------end_max5_patcher-----------
</code></pre>

Live 8.2.2b3, Max 5.1.8, OSX 10.6.4 Intel.

does this seem familiar / reproducable to anyone?

thanks.

[pid]

ok, apologies. the problem seems to be explicitly with pattrstorage in live 8.2.2b3

the problem is with the 'activewritemode' parameter, which i really need.

1st of all, if one changes ANY default attributes to the standard works-with-live pattrstorage setup, pattrstorage stops woking as it should (embedding presets within the device) and starts asking to write a json file to disk on close.

2ndly, if one wants the 'activewritemode' attribute set to '1' by editing the inspector instead, pattrstorage crashes live then max in order. AND, now, not even the command-shift trick works when loading or opening the device, the pattrstorage activewritemode 1 has permanently corrupted the device and i cannot get in to it to fix it and i lost a whole bunch of patching.

any thoughts? / similar experiences? i can provide crash reports if necessary.

thanks.

[hoffman2k]

Pattrstorage got updated in 5.1.7 I think. You need to turn "Intial Enable" and "Auto-update param..." on. Then you won't be asked for a .json file.

To recover your work. Open the device in a text editor. There will be some weird stuff in the first line, the rest is pretty much a normal Max patch which you can copy/paste. Paste it in a new device.

Activewritemode is written in Italic font. That usually means you can't set it in the inspector.
Try this instead [pattrstorage mypattrstrge @activewritemode 1]

[pid]

hi hoffman.

thanks.

yes, i knew all that, and using activewritemode as an attribute to the box also crashes live. and i had "Intial Enable" and "Auto-update param" on already. it was working fantastically until i started messing with the 'activewritemode' attribute.

are you saying that you cannot reproduce this at all? because as soon as i use any attribute, box or inspector, live / max either starts erroneously asking to write .json files or crash.

and as ever, these things happen towards the end of a very complex device build, and scouring the text file makes me weep...

best,
pid

[hoffman2k]

Ah. Didn't realize you had it all set. Some settings are lost when copy/pasting.

I tried the code and can't reproduce the crash.

[jeremydb]

Pattrstorage got updated in 5.1.7 I think. You need to turn "Intial Enable" and "Auto-update param..." on. Then you won't be asked for a .json file.

To recover your work. Open the device in a text editor. There will be some weird stuff in the first line, the rest is pretty much a normal Max patch which you can copy/paste. Paste it in a new device.

Activewritemode is written in Italic font. That usually means you can't set it in the inspector.
Try this instead [pattrstorage mypattrstrge @activewritemode 1]

That means that you have to "freeze" it from the menu on the left of the inspector row in order for it to be saved with the patcher. Otherwise, writing it into the box is the way to go.

Anyway, I can't reproduce using the attached patcher on OSX. Do I need "myabstraction"? Do you have a crashlog?

Jeremy

[hoffman2k]

Pattrstorage got updated in 5.1.7 I think. You need to turn "Intial Enable" and "Auto-update param..." on. Then you won't be asked for a .json file.

To recover your work. Open the device in a text editor. There will be some weird stuff in the first line, the rest is pretty much a normal Max patch which you can copy/paste. Paste it in a new device.

Activewritemode is written in Italic font. That usually means you can't set it in the inspector.
Try this instead [pattrstorage mypattrstrge @activewritemode 1]

That means that you have to "freeze" it from the menu on the left of the inspector row in order for it to be saved with the patcher. Otherwise, writing it into the box is the way to go.

First time I notice there's a damn freeze button there. Add this one to the list of "Did you know...?"

[pid]

one of the reasons i wanted activewritemode 1 was so that a couple of pattr objects in an abstraction were NOT exposed to the main pattrstorage in the parent device AND not saved to my various .json files. whenever i loaded the abstraction the pattr's in question would flip back to active rather than the inactive i had set them as. of course i was being an idiot, and in the end froze those attributes in the pattrs and now i am happy with my parent pattrstorage and .json storing.

i had already frozen the appropriate pattrstorage attributes as i am an attribute-freezing-addict since max 5.0.

of course if i was wanting dynamic control over active this would not work for me as it does crash.

anyway everything is ok for me now, apart from the initial crash. jeremy i will dig out a relavent crash log (there were many) and send it you.

apologies for my verbosity.

p.s. - i presume you were joking with "myabstraction" - i was trying to make an e--j style demonstration patch. i obviously failed.

[jeremydb]

one of the reasons i wanted activewritemode 1 was so that a couple of pattr objects in an abstraction were NOT exposed to the main pattrstorage in the parent device AND not saved to my various .json files. whenever i loaded the abstraction the pattr's in question would flip back to active rather than the inactive i had set them as. of course i was being an idiot, and in the end froze those attributes in the pattrs and now i am happy with my parent pattrstorage and .json storing.

i had already frozen the appropriate pattrstorage attributes as i am an attribute-freezing-addict since max 5.0.

of course if i was wanting dynamic control over active this would not work for me as it does crash.

anyway everything is ok for me now, apart from the initial crash. jeremy i will dig out a relavent crash log (there were many) and send it you.

apologies for my verbosity.

p.s. - i presume you were joking with "myabstraction" - i was trying to make an e--j style demonstration patch. i obviously failed.

Just following up here -- I am going to close this issue out unless you have more information about the crashing.

Jeremy

[pid]

hi jeremy, sorry for lateness of re-reply.

ever since you emailed me the pattrstorage update, then we all updated to 5.1.8, things have been pattr-perfect over here. so thanks thanks thanks. the bernstein does the magic once again.

one OT thing to mention: when i did start using the new pattrstorage in my device, of course it forgot all of it's client mappings, e.g. hours of experimentation with linear/exponential fades, threshold points, and more importantly orderings. luckily i am ocd so i had an image saved of my client window, which i have been in the habit of doing for some time through bad experiences - however one still has to fill everything in again.

am i being total idiot? is there a way round this? is it possibly to save client window state in an m4l device where there is no explicit @savemode? if not, that would be a great feature one day!

but anyway, generally i am superhappy.

best