dat merge using multiple samples! pretty important

Wishlist and RFEs
1

I think this RFE is now quite important, because of the major restructuring of DAT export methodology:

what was
myop/mychan

is now
myopmychan

causing a LOT of extra nodes to be added just to - for example - column replace an exported cell with a user supplied one:

As an example of what I often, very often find my self dealing with (occasionally before, lots of times since adopting the new export format - hence my plea for this RFE):

inside a master, I have a COLORS DAT representing possible colors:

font on 1 1 1 1
font off 0 0 0 0
bg on 0 0 0 0
bg off 1 1 1 1

next to it, I have an IN DAT which represents user overrides:

then a MERGE with COLORS and IN - with IN column replacing COLORS (appending unmatched in other cases where the override is allowing the user to add his/her own exports to parms inside a master).

Actually the above example is not quite representative of the multitude of times when I’m dealing with parameters that need to be exported, but it’s a case I just run into (so no directly related to the change in export format, but a good example for my RFE nonetheless).

so to override one of the above I would connect a DAT to the IN:
font off .5 .5 .5 .5

the problem is replacing the right row, right? Of course it can be done joining/merging/resplitting the columns (so that font_off for example can be properly found then resplit into fontoff), but it’s a great pain, a waste of cycles and memory when duplicated hundreds of times.

How about a merge that uses a start and end index and internally merges/splits the cells?

ie. replace columns start_column=0 end_column=1

would work just fine!!!
how about it? Or… is there a way just as efficient to do it?

2

btw I just implemented the poorman’s version of what I’m asking for (merging based on two columns) and it took 25 DATS!!!

3

Hey Dani, please send an example of this, your input tables and desired output. We are having trouble understanding the challenge you are facing.

4

here’s exactly what I’m talking about

I have a table with something like
fonton1
fontrollover1
bgrollover1

and need to merge it with a table like
fontrollover2

basically I need a replace columns based on matching font AND rollover

see attachment example (with 25 dats…)
base1.tox (2.16 KB)

5

Actually I just noticed there’s a bug in that comp, but you’ll get the idea of what I’m trying to do hopefully.

6

if I understood you correctly, then here’s a version with 13 nodes. I’m sure that it can be reduced to 8 nodes or so with more time for development. replace.toe (5.32 KB)
What I usually do for these cases is change my input format (col0 + col1 in your case) to be just one column using a syntax like
font:rollover…
then you can easily merge replace like you want it. In the end just use a convert DAT and split columns at ‘:’
Basically what I did in the example, but just designed like that from the ground up

but I agree, using replacing by multiple columns would be a nice addition

7

thanks I’ll check it out!

And yes, I never really thought about this RFE until I started implementing the scattered new method for exporting from DAT’s. Before it was

path/op/parmvalue

so it was easy to to replacements.

So changing the format back to the above means I have to resplit the DAT again to reformat it to the proper new export format - which I very much dislike because it’s now more for folks who set those parms by hand than for programmers (because there’s no easy “handle” such as /path/op/parm for a network to look at what is actually inside an export dat, or to replace rows etc.).

more than number of dats and memory I worry about cpu cycles.

d

8

Hey Dani,

3 dats with your old formating converted into the new format via a Convert DAT splitting cells at “/”

Otherwise have a look at how we deal with these things in TUIK (for example buttonmomentary/button), I find it a very clean approach to deal with different panel states.

cheers
Markus
base1.tox (926 Bytes)

9

If you use any path names in these DATs than just replace the “/” with something like %

cheers
Markus