.lift-ranges vs. LIFT

[Piotr Bański]

This is just a request for confirmation/explanation/hints.

I have noticed that the content of .lift-ranges files does not, on the
whole, conform to the LIFT schema. One example that can illustrate that
is the following fragment, where <trait> is added after you enable
inflectional features for nouns, and then export to LIFT. The problem
with that fragment is that <trait> is not licensed in this position in
the LIFT schema -- it may not be a child of <range-element>.

<range-element guid="93bca1bc-f405-49cc-9a6a-ca1d3dd3b117" id="Noun">
<label>
<form lang="en"><text>Noun</text></form>
</label>
<abbrev>
<form lang="en"><text>n</text></form>
</abbrev>
<description>
<form lang="en"><text>A noun is a broad classification of parts of
speech which include substantives and nominals.</text></form>
</description>
<trait name="inflectable-feat" value="nagr"></trait>
</range-element>

Having a schema is good in general, especially when your configuration
files acquire new uses and people like myself start meddling with them
in an attempt to transfer their data smoothly into FLEx. So this message
has actually several possible conclusions: one is "consider adjusting
LIFT", another "consider getting conformant", yet another is that it's
the developers' right to do whatever suits the tool, but perhaps you
could consider deriving a separate schema for .lift-ranges, to partly
document it.

And in general, is the documentation for .lift-ranges available
publicly, please? It's a bit hard to reverse-engineer the modifications
of this file from FLEx exports. I'll be grateful for any pointers.

Best,

Piotr

[John Hatton]

Hi Piotr,

>I have noticed that the content of .lift-ranges files does not, on the
whole, conform to the LIFT schema. One example that can illustrate that is
the following fragment, where <trait> is added after you enable inflectional
features for nouns, and then export to LIFT. The problem with that fragment
is that <trait> is not licensed in this position in the LIFT schema -- it
may not be a child of <range-element>.

I have researched this deeply, but I do see that the LIFT 0.14 documentation
for range-element says it inherits from "extensible", which "allows for
traits and annotational discussions on a range element".

So if the rng schema does not allow for this, it is likely that we need to
fix the schema. Please raise the issue on the LexiconInterchangeFormat
google group; you can submit a patch to the schema there, or just raise the
issue.

Thanks
John Hatton
SIL International Language Software Development, PALASO, and SIL Papua New
Guinea

[Piotr Bański]

Thank you, John.

LIFT 0.14 seems quite ghostly to me -- I'm not sure what uses it. LIFT
0.15 has just got finalized (as far as I was able to gather -- at least
the google docs are frozen now), while FLEx uses version 0.13, or maybe
I should say 0.13.1. And, to be sure, in 0.13, range-element does not
have extensible-content -- it is defined directly as containing
description?|label?|abbrev?, interleaved.

Thanks for the suggestion -- I will post to the LIFT group. I was just
under the impression that this was more of a FLEx issue than a LIFT issue.

Best,

Piotr

[Piotr Bański]

Oh my -- and let me add, for the sake of completeness, that this is on
the basis of 'full-lexicon' LIFT export from version 7.0.6-beta under
Ubuntu.

Apologies for not having mentioned this earlier, in case the info matters!

P.