I think Frank’s work on a test suit is really great, because it is the only
way to avoid a mess of widely diverging CSL implementations. However, what
we need is a way of using a common test suit with all the different
implementations. For example, once I finish the citeproc-hs-php
implementation, I would like to run the test suit to see where citeproc-hs
differs from, say, citeproc-js. At the moment, only Andrea will be able to
change the situation by making changes to the Haskell code base, but at
least we have a transparent way of seeing the differences and implementation
gaps.
Frank, would you consider “outsourcing” the test files (the ones with the
test data >>> …<<<) to a separate repository in the xbiblio svn? Then I
could move forward and write a PHP parser for these files and run the
citeproc-hs processor over them, and other implementations could also do
that.
What do others think?
Christian–
View this message in context: http://n2.nabble.com/Cross-processor-test-suite-tp2746007p2746007.html
Sent from the xbiblio-devel mailing list archive at Nabble.com.
I think Frank’s work on a test suit is really great, because it is the only
way to avoid a mess of widely diverging CSL implementations. However, what
we need is a way of using a common test suit with all the different
implementations. For example, once I finish the citeproc-hs-php
implementation, I would like to run the test suit to see where citeproc-hs
differs from, say, citeproc-js. At the moment, only Andrea will be able to
change the situation by making changes to the Haskell code base, but at
least we have a transparent way of seeing the differences and implementation
gaps.
Frank, would you consider “outsourcing” the test files (the ones with the
test data >>> …<<<) to a separate repository in the xbiblio svn? Then I
could move forward and write a PHP parser for these files and run the
citeproc-hs processor over them, and other implementations could also do
that.
What do others think?
Bruce has asked me to move the code under the csl directory, and I
will do that soon. At the moment the test cases don’t provide a
means of testing locators and positioning hints, which are required
for pinpoint citations. We need to settle the format for delivering
that data, and are hoping to hear views from people involved in the
applications currently using CSL processors (Zotero, Pandoc,
Mendeley). Once the input format for in-cite data has been settled,
I’ll make the changes to the test cases, adapt citeproc-js to be sure
that the test cases can be run, and then move them to their new home
under csl.
Frank