Default track descriptions in are too long

[EricR86]

Original report (BitBucket issue) by Eric Roberts (Bitbucket: ericr86, GitHub: ericr86).

---

As per discussed on the mailing list,

The default track descriptions, which contain a list of the name of every dataset used in a segmentation, are too long, especially when there are more than a couple of tracks. The full name of the track spills out to the left and write.

description="segway ${NUM_LABELS}-label segmentation of ${LIST_OF_TRACKS}" [this goes above the track] name="segway.$UUID" [this is to the left of the track and I believe needs to be unique or there will be difficulties loading multiple tracks]

Maybe we should change the default description to:

• not have the list of tracks
• have "annotation" instead of "segmentation"
• include $IDENTIFYDIR or ($IDENTIFYDIR and $TRAINDIR) as identifiers (as they are specified on the command-line, with relative paths).

Psuedo-template choices:

1) Segway ${NUM_LABELS}-label annotation ($IDENTIFYDIR) 2) Segway model $TRAINDIR annotation $IDENTIFYDIR 3) Segway model $TRAINDIR ($NUM_LABELS labels, $NUM_TRACKS datasets) annotation $IDENTIFYDIR 4) other

We could make the description a Python template that is fed a dictionary with all those options. Would prefer a command line argument to an environment variable though.

[EricR86]

Original comment by Michael Hoffman (Bitbucket: hoffman, GitHub: michaelmhoffman).

---

Let's make the default the equivalent of Segway ${NUM_LABELS}-label annotation ($IDENTIFYDIR).