[sword-devel] xref encoding (Re: The repository with Russian SWORD modules)

[Karl Kleinpaste]

Checking on the state of these modules, the first one I tried in Xiphos
was RSThcsK.  Xiphos crashed on it when I moused over an xref.  I was in
Gen 4:3 when it happened.

Footnotes in RSThcsK are encoded e.g.:
<note type="crossReference">
<reference osisRef="Lev.2.12"></reference>
<reference osisRef="Num.18.12"></reference>
</note>

Compare to NASBnew, where a Gen 4:2 xref is encoded:
<note n="A" osisID="Gen.4.2.xref.A" type="crossReference">
<reference osisRef="Luke.11.50">Luke 11:50</reference>, 
<reference osisRef="Luke.11.51">51</reference>
</note>

The difference, it seems, is just whether there is non-empty content
between <reference></reference>.

The specific failure occurs because Xiphos was not prepared for an empty
return from

getEntryAttributes()["Footnote"]["1"]["RefList"].c_str()

which is how the backend module of Xiphos obtains xref content, where
"1" is the note# within the verse.

Obviously Xiphos should not crash in such circumstance regardless.  But
I need to know what manner of self-defense is in order: Ignore it
silently, pop up a warning "this module is evidently improperly
encoded", announce an engine failure, ...

What I'd like to know is:
- whether RSThcsK's encoding is proper, and
- whether the engine's empty response in reaction to that encoding is
proper.