Details of the construction of DCOR1-changes-only-schemas and FPDAM1-changes-only schemas

[MURATA Makoto (FAMILY Given)]

1. Introduction

This mail describes the way I constructed DCOR1-changes-only-schemas 
and FPDAM1-changes-only schemas.

2. Normalization

I relied on the schema attachment files, say O, in the original 29500
and the schema attachment files, say R, in the DCOR1/FPDAM1.  But some
annoying differences between O and R required normalization.

A) Annotations exist only in O.

As a remedy, I wrote a simple XSLT program for eliminating
annotations.

B) XML declarations

They exist in O, but are absent in R.

C) Indentation or line breaking is different

It appears that different tools were used to format XSD schemas in O
and those in R.  O uses a tab character for indentation while R uses 2
space characters.  Line breaks are also different.

I used oXygen for pretty printing XSD schemas.

D) Comments are different

Some comments exist in O but no such comments exist in R.  I removed them.


3. Creating patch files and then schemas

After overcoming these differences, I performed the following steps.

1) Create patch files from O and R using the diff command.  We can
   obtain a schema in R by applying this patch file to a schema in O.

2) Divide this patch file into two patch files: one is for DCOR1, 
   while the other is FPAM1.  This step is labor intensive, but
   I was able to find some bugs in the DCORs and FPDAMs.

3) Create DCOR1-changes-only schemas and FPDAM1-changes-only schemas
   by applying the patch files.


4. Observations

At present, strict schemas and transitional schemas are completely
separated, although they are very similar.  This almost doubles the
burden of schema maintenance.


Cheers,
Makoto