Large Patches, objects unconnecting/connecting to different objects

Mike S's icon

Now this is a strange one. Most of my work at the moment is done on large patches, plenty of connections, Max and MSP. Occasionally I will open a patch and see something connected to an object that I haven't connected it to, it's often near the object it should be connected to. There are also times when a connection is missing, and I have had no reason to get rid of it and no recollection of getting rid of it. It's happened enough over the years for me to think maybe, just maybe, it's not me thats doing it and is some bizarre glitch in the program.

I know this would be very hard to reproduce, but I wonder if anyone else has noticed it before?

Roman Thilenius's icon

i saw that ONCE, and this should only happen when the file you open is corrupt.

Hugobox's icon

Sometimes the connection is simply invisible (not hidden) and by simply moving the object, the connection is redrawn.
But today I've experienced some objects disconnecting on a patch running 5.1.7.

pdelges's icon

Same thing here, with big patches. Under Max 4.6 (OSX.6.5). Not reproductible.
Roman, you may be right, but I don't understand why suddently many files in this old project are corrupted.

Anyway, adding an simple object in this patch may now have to be followed by hours spent into wrong connections hunting.

seejayjames's icon

I've for sure seen ghost patchcords after changing things, especially opening and closing a subpatch, after moving inlets or outlets around. But a quick lock/unlock has always cleared it up (redraw). The actual connections were never there, it just looked like they were.

pdelges's icon

@seejayjames: the patches are really changed, and so are their functionalities. Patch cords disappear, new connections are made, and even some pictures pasted in the patch disappear. A redraw won't change anything.

seejayjames's icon

Ah, gotcha. It just seemed like they hadn't changed. Though I knew the inlets/outlets changing worked, it's just that the prior patchlines (before swapping around) were still visible, but in reality were no longer "there". The lock/unlock always erased them, that's why I thought it was a redraw of the patch which did it.

jvkr's icon

How large is large (in terms of kb)?
Never experienced this problem with max 5, but first time it happened (max 3 or 4) I realized that when the entire program is contained in a single patch, and when that file gets corrupted, then all work is lost. My strategy is to work with many, many abstractions, that are relatively small in size (10 to few 100 kb). Sorry, this doesn't really help to get your work back.

_
johan

pdelges's icon

@jkvr it's not that large (but cpu intensive and quite complex). The project contains 137 abstractions, their size being smaller than 50KB (besides some patches containing "big" tables).

Don't worry about getting my work back: I have backups on many places... What I lost are just the latests modifications, probably since some files get corrupted.

p

jvkr's icon

Hmm Patrick, sounds like we adopt similar strategies. Is the folder you're working in, included in the search path? Once I had random behavior when such was the case.

_
johan

oootini's icon

i've noticed this in max 5.1.6 on win vista. when i have a large patch and then add objects to presentation, if i have to drag the object a good distance the connections are sometimes lost.

pdelges's icon

@jvkr - no, the folder isn't explicitly in the search path.

This morning, I worked on a brand new patch (included in the same big project as previously - the only one I have that is still under Max4), and now, once again, the patch is totally scramble. Between now and the previous time I saved, there are +/- 20 changes in the patch cords (and again all pictures pasted disappeared).

So I suppose I have to give up and try to work with an older OS for this project :-(

Lars Graugaard's icon

I have had this problem as well with corrupted foles, on 5.1.6 & .7 It is probably related to some jvm issues, discussed in this thread - https://cycling74.com/forums/java-related-crash-2