I’ve created this place holder page.
If you want to add to it, and don’t have permission, let us know.
Just to put my earlier posts in context:
I’m very much for more expressive conditionals. Like most others, it seems, I like the current compact syntax and I am therefore reluctant to add nested elements instead. However, having seen the American Law Style I understand why you would want to add them.
I find that supporting two different syntax variants for conditionals long term adds unnecessary complexity to both specification and implementation. The added overhead is simply not justified by the luxury of simple styles (even though they are the majority) being more compact – actually, for simple conditions the proposed syntax doesn’t become unwieldy exactly.
I would prefer to keep an extended compact syntax; if beasts like the American Law Style leave us no choice but add nested elements to manage their complexity, so be it. In that case I would drop the compact syntax though (even though I do like it). It would be OK for me to not drop it right away, but deprecate it and drop it only with a major release (more on that in a moment) – but I think it should be made very clear that the nested conditional syntax is the new standard.
This is only my opinion, though, and at the end of the day, I think that CSL benefits from the improved conditionals. I do have my reservations stated above – but I would not want these reservations to block the conditionals being added.
As far as versioning is concerned, I realize now that Rintze and I are probably of the same opinion, but that I was not aware that the plan was to drop incompatible legacy features (like the old ordinals) with the next major release – because of the way the spec reads right now I was under the impression you wanted to retain the old ordinal scheme for good:
http://citationstyles.org/downloads/specification.html#ordinal-suffixes
I think the best approach in such cases is to mark the feature as deprecated. So in 1.0.1 it could say that the old ordinal scheme has been deprecated; 1.0.1 processors still must support the 1.0 scheme (preferably, in my opinion, by checking the version number instead of the feature detection used here) and that it will be dropped in the future. The upshot of utilizing the version numbers for this purpose is that we do not need to come up with suitable feature detection mechanisms like in this instance.
I think that there should be a version of the schema for each release available. This is currently the case on GitHub using tags – that works very well. A style and locale should be allowed to specify exactly which version of the spec it is built for – including minor releases. This way a cite processor can adapt its processing routines (e.g., which ordinal algorithm to use) and the test suite knows against which schema to validate the style or locale.
Sylvester
I’m very much for more expressive conditionals. Like most others, it seems, I like the current compact syntax and I am therefore reluctant to add nested elements instead. However, having seen the American Law Style I understand why you would want to add them.
I find that supporting two different syntax variants for conditionals long term adds unnecessary complexity to both specification and implementation. The added overhead is simply not justified by the luxury of simple styles (even though they are the majority) being more compact – actually, for simple conditions the proposed syntax doesn’t become unwieldy exactly.
If we decide to drop the old syntax for conditionals, we could make
simple conditions slightly more compact by optionally forgoing the
cs:conditions element if there is only a single cs:condition, e.g. see
the cs:else-if in:
As far as versioning is concerned, I realize now that Rintze and I are probably of the same opinion, but that I was not aware that the plan was to drop incompatible legacy features (like the old ordinals) with the next major release – because of the way the spec reads right now I was under the impression you wanted to retain the old ordinal scheme for good:
I probably could have been more explicit in the spec. But the new
ordinal syntax is better in all respects than the old one, so the only
reason it’s still around is for legacy reasons.
I think that there should be a version of the schema for each release available. This is currently the case on GitHub using tags – that works very well. A style and locale should be allowed to specify exactly which version of the spec it is built for – including minor releases. This way a cite processor can adapt its processing routines (e.g., which ordinal algorithm to use) and the test suite knows against which schema to validate the style or locale.
I actually proposed something like this a while back:
Rintze