View previous topic :: View next topic |
Author |
Message |
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Tue Mar 30, 2004 11:50 pm Post subject: |
|
|
!! Woah! It all links! Good going! Have you uploaded the update to subversion yet? Please do! I can't wait.
Thanks,
John |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Tue Mar 30, 2004 11:57 pm Post subject: |
|
|
Indeed, send the code and I'll post it for all to see. _________________ I really like the vest! |
|
Back to top |
|
|
evetsyor
Joined: 27 Mar 2004 Posts: 12
|
Posted: Wed Mar 31, 2004 7:07 am Post subject: |
|
|
the file only content the headers and the classes, i want to make the whole kit to compile with only the declaration, the implementation is all commented out.
then if the interface can compile, then i can start putting the real code ins.
- Steve |
|
Back to top |
|
|
evetsyor
Joined: 27 Mar 2004 Posts: 12
|
Posted: Wed Mar 31, 2004 7:14 am Post subject: |
|
|
and also subversion version is the DWT M6 version, my branch is from M8 version, and i did, cuz i didnt wanted to modify the DWT one, i started clean to keep the original import in the same order etc.. once ive break all the problem we can have with the port, then i figure out with Brad about what we do, either i try to re-make the DWT to work like mine is (now that ill know how to) or merge DWT into my branch, it doesnt matter much for now. all i do is moslty a test for now. trying to find solution to all problem the porting can have and fix them before doing the real work.
- Steve |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Wed Mar 31, 2004 4:20 pm Post subject: |
|
|
evetsyor wrote: | the file only content the headers and the classes, i want to make the whole kit to compile with only the declaration, the implementation is all commented out.
then if the interface can compile, then i can start putting the real code ins.
|
That's actually a very smart way to do it since a lot of the issues were initially bound up in the class inheritances and interdependencies. If you can get throught that, it's a huge start. The only problem with this is that it still circumvents secondary issues such as contained classes (objects declared within a class). These have been bringing up there own forward reference issues. But of course, it's still a good idea to start with the primary problem as you have done.
I may fiddle around with the subversioned DWT source to see what the new 0.82 compiler does for it. Commenting out the implementation details is valid idea here too.
Keep us posted,
John |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Sat Apr 17, 2004 9:56 pm Post subject: Compilation process |
|
|
I'm curious to know how the DWT "non-implemented" class/interface compilation process is going. If it has succeeded, it's best that we use this as a base for improvement of DWT and start filling in the classes soon.
Steve, do you have a status? I'd really like to start working on implementing some of the code soon. |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Sat Apr 17, 2004 9:59 pm Post subject: |
|
|
I got a mail from Steve saying he was only doing the non-implemented classes to see if he could get around the forward references. He got pretty far, but ran into it again.
We left it that his work was not going to be folded in, and we should resume work on the Subversion tree.
That's what I'm going to do. John, if it's you and me, let's lay out a plan. There are a bunch of instanceof replacements as well as delegate work to be done. Then we have the overall "get the thing to compile" task that we'll both be working toward.
Thoughts? _________________ I really like the vest! |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Sat Apr 17, 2004 10:27 pm Post subject: |
|
|
brad wrote: | I got a mail from Steve saying he was only doing the non-implemented classes to see if he could get around the forward references. He got pretty far, but ran into it again.
|
That's too bad! I feel this is one of the most important initial issues that must be resolved before any real headway can be made. If the skeleton is not set up and organized is such a way to work, than it seems futile to do a huge implementation task first. It's almost like saying lets build the brick house now.... we'll lay the concrete foundation later . The forward reference issues were the main thing that discouraged me earlier. Really, I feel if we can't succeed in getting those issues fixed, then I'm going to be very leary of investing tons of time implementing something that has no guarantee of working in the future (and it'll be harder to make work then too).
brad wrote: | That's what I'm going to do. John, if it's you and me, let's lay out a plan. There are a bunch of instanceof replacements as well as delegate work to be done. Then we have the overall "get the thing to compile" task that we'll both be working toward.
|
There's actually quite a few internal things that we will have to decide on what we are going to do about. We can pick Ant's brain about delegates. Instanceof, I think we will be able to fix in one of two standard ways.
But I still believe that getting the class/interfaces ordered properly is an important first consideration. Someone also suggested the use of interfaces to get around the issue, but that may create too big of a departure from the original design for my liking.
The advantage to getting the basics compilable to compilation first:
1) gradual implementation following will be more testable in stages
2) ability to test more often
3) easier to diverge source in select areas
...plus a few other pluses I can't think of at the moment
Your thoughts? Anybody's thoughts?
Later,
John |
|
Back to top |
|
|
jcc7
Joined: 22 Feb 2004 Posts: 657 Location: Muskogee, OK, USA
|
Posted: Sat Apr 17, 2004 11:08 pm Post subject: |
|
|
JJR wrote: | That's too bad! I feel this is one of the most important initial issues that must be resolved before any real headway can be made. If the skeleton is not set up and organized is such a way to work, than it seems futile to do a huge implementation task first. | I'm just a fly on the wall with respect to DWT, but I think you're right. If the unimplemented classes don't compile, how can we expect to get implemented classes to compile?
Would it be possible to check Steve's work into Subversion as a subdirectory of DWT (call it "framework" or "test")? Then as Steve and/or other interested parties make progress at getting parts of the framework to compile, they can commit the better versions. Whether they get folded in or not doesn't necessarily matter, but at least the framework could be used as a guide to successful future compilation.
This forward reference problem seems to be a weakness of D, and if we don't isolate the trouble spot earlier, then we shouldn't be surprised if they pop up later.
The other implementation details could still be discussed now, but I don't see any cause to give up on the framework.
But I'm still just on the outside looking in... |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Sun Apr 18, 2004 12:06 am Post subject: |
|
|
Here's what I think... an opinion... or perhaps it could be called a compromise.
The task of arranging the classes/interfaces in a workable skeleton system does not necessarily have to be done by two people. I wouldn't mind taking on that task; I still think it needs to be done at the start of the project. Perhaps we could indeed have a look at what Steve did, and that may accelerate this thing even more (rather than repeat all that work).
Brad, if you want to start discussing about and working on implementation issues, that would be great too... Especially getting the discussions started early would be very important in itself. I guess you already started a little of that on delegates. Hopefully we could begin recruiting some help on dwt too.
With this arrangement, at least we wouldn't be overlapping our labor. As soon as I finished (heh!) with my task, I'd jump head first into actual implementation.
I'll probably ask here in parallel for adice on class interactions/schemes for getting the compilation process working as I run into problems (which I recall all to vividly ).
Any more ideas? Comments? Does this sound feasible? Or am I out to lunch?
Later guys,
John |
|
Back to top |
|
|
brad Site Admin
Joined: 22 Feb 2004 Posts: 490 Location: Atlanta, GA USA
|
Posted: Sun Apr 18, 2004 12:13 am Post subject: |
|
|
I'm fine with these suggestions. And you divided it up appropriately. I'd like to get to the implementations, while you work on the framework. I can always cut and paste a file at a time to see if we can compile it.
Here's the issue I see, though. Both parties will get to the same place. When you're done with framework, you should get it to compile. The minute you try to implement even one file, it will bomb because you need all of the dependencies. Same on my side. If I don't implement everything correctly, without bugs, then we're in forward references hell.
Since we'll end up in the same place, I say we get after it, and not wait for Steve's code. Another question: we've been converting 3.0M6. Do we keep going and if we get it working, do a diff on the java code? Or do we switch to 3.0M8? My vote is for M6, keep going, then do the diff. _________________ I really like the vest! |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Sun Apr 18, 2004 12:31 am Post subject: |
|
|
brad wrote: | I'm fine with these suggestions. And you divided it up appropriately. I'd like to get to the implementations, while you work on the framework. I can always cut and paste a file at a time to see if we can compile it. |
Ok... cut and paste may be a possibility but it will certainly bomb like you say below.
brad wrote: |
Here's the issue I see, though. Both parties will get to the same place. When you're done with framework, you should get it to compile. The minute you try to implement even one file, it will bomb because you need all of the dependencies. Same on my side. If I don't implement everything correctly, without bugs, then we're in forward references hell.
|
True, this compromise may make for some fairly significant, yet unavoidable issues. I don't see wholesale cutting and pasting as immediately workable. The immediate benefit of your implementation task will be in solving D related conversion issues before they are integrated. The overall worakability of the implementation of a class as a whole becomes a whole different matter, something we'll have to definitely be working together on. If you can at least make sure some of the lower level issues are valid D, that will be a start. I agree with you that it will not go smoothly, though. I'm hoping that this way, at least, we can gradually add pieces (empty methods as they are needed). But then again it's a whole new game of satisfying more and more and more dependencies to make things work....
brad wrote: | Since we'll end up in the same place, I say we get after it, and not wait for Steve's code. Another question: we've been converting 3.0M6. Do we keep going and if we get it working, do a diff on the java code? Or do we switch to 3.0M8? My vote is for M6, keep going, then do the diff. |
Agreed. M6 is fine.... I see a lot of D conversions being very necessary anyway. |
|
Back to top |
|
|
JJR
Joined: 22 Feb 2004 Posts: 1104
|
Posted: Sun Apr 18, 2004 12:35 am Post subject: |
|
|
Have there been any recent changes to DWT committed by anyone to subversion? |
|
Back to top |
|
|
jcc7
Joined: 22 Feb 2004 Posts: 657 Location: Muskogee, OK, USA
|
Posted: Sun Apr 18, 2004 1:42 am Post subject: |
|
|
It looks like the latest revision is when brad added a blank line to HelloWorld.d on 3/21/2004 (revision 13). |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|