[sword-devel] ERV issues in And Bible

[David Haslam]

Experimenting with changes to your OSIS XML file, to see if I can influence
the number of duplicates:

*First result: *

After I removed every line break element <lb /> and rebuilt the module,
the number of duplicate div elements was reduced to 357.

This suggests that we may be able to devise a workaround as a holding
measure until osis2mod can be fixed.

It's significant in that all the previous occurrences associated with the
Psalm titles were eliminated.

Looking closer at the results of my filter  to "Extract and count milestone
div elements",
it's evident that we have more than a mere duplication problem!

After I sorted the lines in descending order, the first few lines show that
we have a "multiplication" issue, which is far worse than mere duplication!

Here are the first few lines after the sort.

00072	<div type="x-milestone" subType="x-preverse" sID="pv1649"/>
00072	<div type="x-milestone" subType="x-preverse" eID="pv1649"/>
00007	<div type="x-milestone" subType="x-preverse" sID="pv12529"/>
00007	<div type="x-milestone" subType="x-preverse" eID="pv12529"/>
00006	<div type="x-milestone" subType="x-preverse" sID="pv2683"/>
00006	<div type="x-milestone" subType="x-preverse" eID="pv2683"/>
00005	<div type="x-milestone" subType="x-preverse" sID="pv5210"/>
00005	<div type="x-milestone" subType="x-preverse" sID="pv4769"/>
00005	<div type="x-milestone" subType="x-preverse" sID="pv12535"/>
00005	<div type="x-milestone" subType="x-preverse" eID="pv5210"/>
00005	<div type="x-milestone" subType="x-preverse" eID="pv4769"/>
00005	<div type="x-milestone" subType="x-preverse" eID="pv12535"/>
00004	<div type="x-milestone" subType="x-preverse" sID="pv74"/>
00004	<div type="x-milestone" subType="x-preverse" sID="pv518"/>
00004	<div type="x-milestone" subType="x-preverse" sID="pv2491"/>
00004	<div type="x-milestone" subType="x-preverse" eID="pv74"/>
00004	<div type="x-milestone" subType="x-preverse" eID="pv518"/>
00004	<div type="x-milestone" subType="x-preverse" eID="pv2491"/>
....

This means that the part of the code than generates the /serial numbers/ for
*?ID="pv#####"*  must be seriously delinquent.

A better measure of the improvement is the increased number of *unique*
milestone div elements.

Previously there were 32461 with count=00001
Afterwards there are 33269 with count=00001, an increase of 808.

Understand that the difference between *sID* and *eID* makes the extracted
lines unique even when they have the same serial *pv* number.

Can any of the SWORD developers shed light on how the serial numbers are
generated?

Best regards,

David





--
View this message in context: http://sword-dev.350566.n4.nabble.com/ERV-issues-in-And-Bible-tp4656923p4656939.html
Sent from the SWORD Dev mailing list archive at Nabble.com.