Erik,
Here is the misunderstanding in a nut shell.
The commit I got so upset about was mainly knee jerk but is served as
a warning to me. The ONLY time I have a problem with any type of
"commit then review" process is when the changes or refactoring have
to do with the frameworks (existing) architecture.
I could care less about meddling in your affairs with everything else,
I understand that is development.
Does this make sense?
Can we get back on the level playing field now that we have always been on?
Mike
Quoting Erik de Bruin <e...@ixsoftware.nl>:
Mike, I'm confused. I'm sure it's me (being a foreigner and all), but
I don't understand what you're asking of me...
I did a big commit 'solo', it nearly was vetoed. The suggestion was I
talk about what I plan to change before actually committing next time
I needed to make changes that might influence other's code. I did
(this thread), but now you seem to be asking me to discuss what I'm
going to do even BEFORE I actually write code, locally?
I'm not sure what your process is, but mine generally starts with a
goal ("enable js output from MXML"), after which if tinker with the
code until it works. This may or may not involve dead ends, reverts or
do-overs. Mostly, what I thought might work doesn't and what ends up
working is not at all what I though it might be. When the code works,
I clean it up, re-format it, run the tests one more time and commit.
I'm not sure how I can discuss changes to the code before I touch the
code. I can, however, discuss what I'll be working on, which I thought
I did...
As the original contributor of the FalconJx code, in my mind you are
the de-facto project lead. I therefore defer to your suggestions, most
of the time ;-) I don't mind that at all, as long as we work as a
team. I'm trying to understand what you think is the best way to
cooperate and how I can best fit that into my work. Please be patient
and maybe explain things "like I'm a 5 year old", just so I understand
what it is you're expecting of me.
Thanks,
EdB
On Tue, Mar 5, 2013 at 10:56 AM, Michael Schmalle
<apa...@teotigraphix.com> wrote:
We did. :)
I just wanted to see if you were reading every word I write. :)
Mike
Quoting Erik de Bruin <e...@ixsoftware.nl>:
It's re-renamed (de-named?).
About 'common', I tried to explain that might be a misnomer due to me
not being a native English speaker.
As stated before, I complete stand behind what you say about moving
everything (as, js and mxml) into one 'codegen', 'driver' and
'visitor' package. I just thought we had agreed to postpone such a
major refactor until some point in the future?
EdB
On Tue, Mar 5, 2013 at 1:16 AM, Michael Schmalle
<apa...@teotigraphix.com> wrote:
Erik;
renamed IASNodeStrategy to INodeStrategy
I disagree, please rename that interface back to IASNodeStrategy.
The only method it has is handle(IASNode node), notice the IASNode. It is
a
IASNode handler strategy.
Can we please be a little more pragmatic at this refactoring and
renaming? I
don't understand what compelled you to want to rename that interface.
I'm really not liking this 'common' folder at all. I really believe
common
API belongs in it's own package, not sub packages of a common directory.
Look at how the falcon framework is laid out, they do not abuse the
common
directory.
Putting codegen and things on a common directory when there is already a
codegen directory is redundant and confusing for others in the future.
That
being said, I said it was MY mistake not making a codegen and driver
directory in compiler. If you want to refactor, do it right and make a
codegen, driver in the compiler, then move the 'as', 'js' and 'mxml' into
the codegen package and axe the common package.
Mike
Quoting Erik de Bruin <e...@ixsoftware.nl>:
Mike et al.,
I have a reasonably big commit lined up. To make AS embedded in MXML
work without doing duplicate work, I figured I could best use the
existing ASEmitter and subclases. To make this work, I needed to add
an ASBlockWalker to the MXMLBlockWalker and make adjustments to some
existing code (refactoring of interfaces and method signatures,
mostly). I was able to keep most of this trickery limited to MXML
classes, but I needed to make some changes to these 'common' and AS
classes:
- renamed IASNodeStrategy to INodeStrategy, as it is now 'common' and
used by both AS and MXML; this renaming touches 'a few' other classes,
like IJSEmitter and the classes in
'org.apache.flex.compiler.internal.as.codegen'
- created IBlockVisitor and IBlockWalker as 'common' interfaces
- refactored IASBlockVisitor and IASBlockWalker to extend these new
interfaces
All tests pass (I even managed to get a few more done for FlexJS) and
the road ahead seems clear...
Let me know if any of this will break anything beyond repair - or at
least beyond a little time spend adjusting code to the new - if I
commit these changes,
EdB
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
--
Ix Multimedia Software
Jan Luykenstraat 27
3521 VB Utrecht
T. 06-51952295
I. www.ixsoftware.nl
--
Michael Schmalle - Teoti Graphix, LLC
http://www.teotigraphix.com
http://blog.teotigraphix.com
