On Thu, Jun 4, 2015 at 2:11 PM, Alex Harui <aha...@adobe.com> wrote: > > > On 6/4/15, 10:31 AM, "Michael Schmalle" <teotigraphix...@gmail.com> wrote: > >>If you mean the AS package for the classes in these SWCs, I have to admit > >> I was a bit surprised to see in the AS output example that Document in a > >> “dom” package. I would think these two SWCs would not have their > >>classes > >> in any AS package (IOW, “package {“). Aren’t classes like Document > >> effectively global just like Math? > >> > >> > >Ok, yeah that was ME, what about naming collisions? If they are in a > >package during AS dev, there is no possible naming collisions with real AS > >classes that might be local or root. If we put them in a package, the > >compiler can boil them down to global during compile, this was the > >JavaScript tags job. > > > > Do you mean a collision with some customer’s Document class in the > top-level package? Wouldn’t that collision eventually cause something to > break when the customer tries to run the output JS? Seems like getting > AmbiguousReference would be a good thing. > >
DUH! It's the simple things I miss. > > > > > >> >2. I am using Rhino and QDox(hack for getting an easy parser for the > >> >jsdocs), so I guess a download jar entry is going to have to be added > >>to > >> >the build script? I am not good at ant. > >> > >> Rhino 1.74 or later is MPL. If we don’t need an earlier version then > >>yes > >> we can download it and Qdox (which is AL). Just check in code that uses > >> these jars (but not the jars) and I’ll get the build scripts to download > >> them. > >> > >> > >I am using the Rhino source code right now but the license says; > > > > * This Source Code Form is subject to the terms of the Mozilla Public > > * License, v. 2.0. If a copy of the MPL was not distributed with this > > * file, You can obtain one at http://mozilla.org/MPL/2.0/. > > Apache prefers to work with MPL from jars so I’ll pull a Rhino jar from > somewhere. > > OK > >> > >> >4. I think I am going to utilize that JavaScript metadata tag for > >>these so > >> >it's easy to resolve stuff during compile. > >> > >> I guess that’s ok with me. Why do you not want to introduce new > >>metadata > >> keywords and/or use asdoc similar to how goog is using jsdoc? I’d worry > >> about a JavaScript metadata statement getting longer and longer as we > >>add > >> more attributes to it. Having each attribute be a keyword or > >>asdoc/jsdoc > >> tag seems like it would be more manageable? > >> > > > > > >Yeah, I can do it through comments, but right now the damn ASDocTokenizer > >is broken, so I don't have an nice clean way to parse the doc comments. > > Hmm. Maybe I should spend some time looking into fixing ASDocTokenizer? > Was the problem that it didn’t work on every AS file we current have? > It doesn't work on anything, there is an infinite loop in the scanner that is created by JFlex, the RawASDocTokenizer is broken. What is weird is I was using the SAME code base when I wrote the asdoc documenter I have 2 years ago and it worked fine. Mike > > -Alex > > >