I will be doing some development in Squeak 3.9 / Seaside (I have played with
Squeak before, this time it will be RealWork). I like the idea of using traits, but have no actual hand-on usage experience. Should I use traits in my work? Know of any real deployed projects that have used traits to good effect? Do the tools manage the cost of all the finer-grained bits, debugging, etc? Are there any best practices I could look at? Will using traits affect (a) possible port to VW? (b) possible use of Gemstone/S? Thanks! _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
2007/9/14, itsme213 <[hidden email]>:
> I will be doing some development in Squeak 3.9 / Seaside (I have played with > Squeak before, this time it will be RealWork). I like the idea of using > traits, but have no actual hand-on usage experience. > > Should I use traits in my work? What is your work? > Know of any real deployed projects that have used traits to good effect? Only on a very small base, 2 or 3 traits. > Do the tools manage the cost of all the finer-grained bits, debugging, etc? Debugging is not a problem. The problem is that the code browser and Monticello are only 80% there, RB is not trait aware, .... > Are there any best practices I could look at? > Will using traits affect (a) possible port to VW? (b) possible use of > Gemstone/S? You mean because they don't support traits? Yes of course. But you can always flatten the traits. Then you'll "only" have to deal with the real portability problems which are big enough. Cheers Philippe > Thanks! > > > > _______________________________________________ > Seaside mailing list > [hidden email] > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
Sounds like you should also consider Sport if portability is a real
possibility in the future.... http://groups.google.com/group/comp.lang.smalltalk/browse_thread/thread/5eba38351042f67e/0f35dbacbdf8f47d?lnk=raot ]{evin Philippe Marschall wrote: > 2007/9/14, itsme213 <[hidden email]>: > >> I will be doing some development in Squeak 3.9 / Seaside (I have played with >> Squeak before, this time it will be RealWork). I like the idea of using >> traits, but have no actual hand-on usage experience. >> >> Should I use traits in my work? >> > > What is your work? > > >> Know of any real deployed projects that have used traits to good effect? >> > > Only on a very small base, 2 or 3 traits. > > >> Do the tools manage the cost of all the finer-grained bits, debugging, etc? >> > > Debugging is not a problem. The problem is that the code browser and > Monticello are only 80% there, RB is not trait aware, .... > > >> Are there any best practices I could look at? >> Will using traits affect (a) possible port to VW? (b) possible use of >> Gemstone/S? >> > > You mean because they don't support traits? Yes of course. But you can > always flatten the traits. Then you'll "only" have to deal with the > real portability problems which are big enough. > > Cheers > Philippe > > >> Thanks! >> >> >> >> _______________________________________________ >> Seaside mailing list >> [hidden email] >> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >> >> > _______________________________________________ > Seaside mailing list > [hidden email] > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > > _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
In reply to this post by Philippe Marschall
>> Should I use traits in my work?
> > What is your work? A tool for software architects, in which I think I see several places where trait-like structuring could be useful. >> Know of any real deployed projects that have used traits to good effect? > > Only on a very small base, 2 or 3 traits. That is quite useful to know, though I was hoping otherwise. >> Do the tools manage the cost of all the finer-grained bits, debugging, >> etc? > > Debugging is not a problem. The problem is that the code browser and > Monticello are only 80% there, RB is not trait aware, .... Ah, also most useful to know. I wanted to use traits more than build up the tooling for it. >> Will using traits affect (a) possible port to VW? (b) possible use of >> Gemstone/S? > > You mean because they don't support traits? Yes of course. All in all sounds like it may not be quite ready yet. Thank you very much for the input.. _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
2007/9/14, itsme213 <[hidden email]>:
> >> Should I use traits in my work? > > > > What is your work? > > A tool for software architects, in which I think I see several places where > trait-like structuring could be useful. > > >> Know of any real deployed projects that have used traits to good effect? > > > > Only on a very small base, 2 or 3 traits. > > That is quite useful to know, though I was hoping otherwise. > > >> Do the tools manage the cost of all the finer-grained bits, debugging, > >> etc? > > > > Debugging is not a problem. The problem is that the code browser and > > Monticello are only 80% there, RB is not trait aware, .... > > Ah, also most useful to know. I wanted to use traits more than build up the > tooling for it. > > >> Will using traits affect (a) possible port to VW? (b) possible use of > >> Gemstone/S? > > > > You mean because they don't support traits? Yes of course. > > All in all sounds like it may not be quite ready yet. Thank you very much > for the input.. Besides the tool issues there are other issues as well. There is somewhere a bug in ClassBuilder and every trait in the image ends up with each user twice. Also copy class doesn't work correctly for traits. It's likely that there are more bugs like this. Cheers Philippe _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
On Sep 15, 2007, at 08:29 , Philippe Marschall wrote: > > Besides the tool issues there are other issues as well. There is > somewhere a bug in ClassBuilder and every trait in the image ends up > with each user twice. That's wrong. New traits don't have multiple users. There are existing traits in the image that have multiple users. They must have got there during the 3.9 process. How that happened has not been found out (as Philippe noted, its potentially a bug in the ClassBuilder). As said earlier, rehashing those sets gets rid of the duplicates. > Also copy class doesn't work correctly for > traits. The problem here is that also methods from traits are copied, but like the bug above, its not a show stopper. > It's likely that there are more bugs like this. Right, but we woun't find them if we wait using traits. In my oppinion, the implementation of traits is stable enough for real use. I mean, part of the Squeak kernel is implemented with traits... Rather a limiting factor is UI support. Although various pleople have been working traits aware tools and extensions, there is missing a coherent and maintained package. Adrian _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
2007/9/17, Adrian Lienhard <[hidden email]>:
> > On Sep 15, 2007, at 08:29 , Philippe Marschall wrote: > > > > Besides the tool issues there are other issues as well. There is > > somewhere a bug in ClassBuilder and every trait in the image ends up > > with each user twice. > > That's wrong. New traits don't have multiple users. There are > existing traits in the image that have multiple users. They must have > got there during the 3.9 process. How that happened has not been > found out (as Philippe noted, its potentially a bug in the > ClassBuilder). As said earlier, rehashing those sets gets rid of the > duplicates. > > > Also copy class doesn't work correctly for > > traits. > > The problem here is that also methods from traits are copied, but > like the bug above, its not a show stopper. > > > It's likely that there are more bugs like this. > > Right, but we woun't find them if we wait using traits. It's just the question whether you want to be the one who spots them in your production project and wait for a fix. Philippe > In my > oppinion, the implementation of traits is stable enough for real use. > I mean, part of the Squeak kernel is implemented with traits... > Rather a limiting factor is UI support. Although various pleople have > been working traits aware tools and extensions, there is missing a > coherent and maintained package. > > Adrian > _______________________________________________ > Seaside mailing list > [hidden email] > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
But philippe none of the "bugs" you mentioned is a runtime problem.
At runtime traits do not exist. Damien worked 5 months full time with traits and everything worked well. So using traits is the only way to make them real. I will try to allocate some times, investigating the problems. The tools are improving too but with no user then people doing them will be Stef On 17 sept. 07, at 09:15, Philippe Marschall wrote: > 2007/9/17, Adrian Lienhard <[hidden email]>: >> >> On Sep 15, 2007, at 08:29 , Philippe Marschall wrote: >>> >>> Besides the tool issues there are other issues as well. There is >>> somewhere a bug in ClassBuilder and every trait in the image ends up >>> with each user twice. >> >> That's wrong. New traits don't have multiple users. There are >> existing traits in the image that have multiple users. They must have >> got there during the 3.9 process. How that happened has not been >> found out (as Philippe noted, its potentially a bug in the >> ClassBuilder). As said earlier, rehashing those sets gets rid of the >> duplicates. >> >>> Also copy class doesn't work correctly for >>> traits. >> >> The problem here is that also methods from traits are copied, but >> like the bug above, its not a show stopper. >> >>> It's likely that there are more bugs like this. >> >> Right, but we woun't find them if we wait using traits. > > It's just the question whether you want to be the one who spots them > in your production project and wait for a fix. > > Philippe > >> In my >> oppinion, the implementation of traits is stable enough for real use. >> I mean, part of the Squeak kernel is implemented with traits... >> Rather a limiting factor is UI support. Although various pleople have >> been working traits aware tools and extensions, there is missing a >> coherent and maintained package. >> >> Adrian >> _______________________________________________ >> Seaside mailing list >> [hidden email] >> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >> > _______________________________________________ > Seaside mailing list > [hidden email] > http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside > _______________________________________________ Seaside mailing list [hidden email] http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside |
Free forum by Nabble | Edit this page |