remove dependency to neurondb.xml

[jdcourcol]

This file is a BBP internal format. It may also contains informations that are unrelated to bluepymm such as repair instructions.
We should have a much simpler format for the end user.

[liesbethvanherpe]

Currently, the "neurondb.xml" input is parsed into a list of (name, fullmtype, mtype, msubtype, layer)-tuples.

[wvangeit]

for this we should solve the file-extension issue. Problem is that at the moment we don't have any info about the file extensions, preventing the use of swc files.

[jdcourcol]

do we need to have the file extension in the file ? can we just have a list of tuple (name, layer, mtype, msubtype) ?

[wvangeit]

Doesn't it really matter how we get it, but the morph release should communicate to us the file extension in some way.
But to me this input file seems to be the right location for that.
I wouldn't be surprised e.g. if hippocampus could have mix of swc and asc in same release

[jdcourcol]

What we want to express here are the morphological properties for each morphology. This is independent of the representation of the morphology (.asc, .swc..).

I would rather have additional parameters that give us the source (input directory, neuroinformatics URL later) and one that give us the preferred representation (.asc, .swc...)

To sum up, I prefer to separate the 2 concerns: morphology properties and how we load the morphology

[wvangeit]

Ok, so it means we need another input file with the extensions of every morphology.

[jdcourcol]

no, just a parameter that says which extension which should use. and we can default to '.asc' so that it is optional in most case

[wvangeit]

and how do we do that with mixed morphologies ?

[liesbethvanherpe]

Latest update:

[
{
"morphname": "morph1",
"dirname": ".",
"extension": "asc",
"mtype": "mtype1",
"layer": "1"
},
{
"morphname": "morph2",
"extension": "asc",
"mtype": "mtype2",
"layer": "2"
}
]

where "dirname" is optional with "." as a default value.

[liesbethvanherpe]

See #139