1) Doing processing on field-level not form-level. 2) This allows us to have a field parametrized by its type. Most fields will be of string type, but upload field will give WUpload.file. 3) This will allow us to nicely deal with choice fields, where we can choose from a list of values of an arbitrary type (in a type-safe way).
Ok, it seems that both forms with and without file upload fields work and work coherently. Also the API seems to be OKish (guys, comments/complains welcome). More testing needed!!
Most of the functionality that needed to be provided by the application now is in stdlib (based on implementation on opalang.org).
…on the way)
when an inner function is contained in a local annotated function
… one-use channels
…lly after renaming it appears to somehow raise assertions later when you had db default values in several (non-linking) packages
Removing the inclusion of opa-plugin-builder.adoc and opa-plugin-browser.adoc caused xmllint to whine about it. One might think about a patch that does not break install_release. :-) Oh, I also removed the warnings generated by asciidoc(1).
* add an env var OCAML_WORD_SIZE for ppdebug. * next step: patch the code at critical sections using ppdebug OCAML_WORD_SIZE 32. * note that this is a mandatory patch on the road to have a 32bit version, but the code needs to be patched. next commits will be to use the env var OCAML_WORD_SIZE in some places for getting things working.
… the first tab if open at generation
…callbacks. Designed to fight against memory leaks.