Closed shaunlebron closed 9 years ago
working on this here: https://github.com/cljsinfo/cljs-api-docs/tree/multi-pass
removing the single version feature to simplify things. its use-case of fast development is covered by the :skip-parse?
and :skip-pages?
options.
I really want the referenced symbols in cljsdoc files to be validated against actual symbols. the original api-docs repo solved this by checking against the canonical symbol list in the separate api-refs repo. but now the repos have merged so that the generated pages have our manual docs.
so, we need a multi-pass system so that the manual docs can depend on the fully generated data as before.
For this, we need preliminary output directories, we'll call
.cljs-api-cache/<version>
, which will be used to incrementally build up intermediate results for each version before creating repo commits.maybe we should use
cljs-api.no-cljsdoc.edn
for data that doesn't have cljsdoc info yet..cljs-api-cache/<version>/cljs-api.no-cljsdoc.edn
.cljs-api-cache/cljsdoc.edn
for caching.cljs-api-cache/<version>/cljs-api.no-cljsdoc.edn
.cljs-api-cache/<version>/cljs-api.edn
This will also allow a faster CI process (generate all data, but only generate the most recent pages).
Need to add new options to cmd param:
:task
:cljsdoc
:docset
nil
:version
<#>
<version>
:latest
:latest
:catalog?
true
false
<#>
false
:skip-pages?
true
false
false
:skip-parse?
true
false
false