Is service construction really so expensive? After all, we're doing exactly that for page/component reloading (throw away all loaded pages/components if a change to any of them occur), and I suspect that pages and components are considerably more expensive to construct than services? Given that service reloading will primarily be occurring in development environments, this seems like it may be a premature optimization. Or is there some reason other than performance that you don't want to have to throw away the other services?
Robert On Aug 11, 2010, at 8/114:04 PM , Howard Lewis Ship wrote: > Isolation. If one service impl changes, I don't want to have to throw > away all the other service implementations that didn't. > > On Wed, Aug 11, 2010 at 2:01 PM, Igor Drobiazko > <igor.drobia...@gmail.com> wrote: >> Why actually creating a class loader per service? Why not creating a single >> class loader which is used to load all classes? This service can be thrown >> away if one of the classes has been changed. This would solve the problem of >> class and package identity. >> >> On Wed, Aug 11, 2010 at 10:20 PM, Howard Lewis Ship <hls...@gmail.com>wrote: >> >>> Perhaps there's a way we can analyze the service and determine that it >>> is not compatible with live service implementation reloading. >>> However, part of this should be addressed by an existing bug, which >>> indicates that Tapestry should automatically load the base class of a >>> service implementation class in the same class loader. >>> >>> On Wed, Aug 11, 2010 at 1:07 PM, Kalle Korhonen >>> <kalle.o.korho...@gmail.com> wrote: >>>> On Wed, Aug 11, 2010 at 12:32 PM, Thiago H. de Paula Figueiredo >>>> <thiag...@gmail.com> wrote: >>>>> In which package the proxy is created? If it's just a matter or putting >>> them >>>>> in the same package, just name the proxy ${className}___TapestryIoCProxy >>> or >>>>> something like that. Is this possible? >>>> >>>> Do read what Howard's saying. Classes in Java are only uniquely >>>> identified by the fully qualified class name and the loading >>>> classloader. I wonder if providing a global flag for controlling the >>>> service reloading and .enableReloading() would make matters worse? >>>> Probably, more configuration typically causes only more mess. What >>>> about disabling service reloading automatically if there's even one >>>> protected/package-private operation and logging a warning? >>>> >>>> I doubt that common abstract classes for services are that typical. >>>> I'd personally live with the restrictions live service reloading >>>> imposes but I can see how it might make many others unhappy. Had this >>>> been in place from the beginning, I doubt it would have caused much of >>>> a fuzz, but now it just might. Damned if you do, damned if you >>>> don't... >>>> >>>> Kalle >>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: dev-unsubscr...@tapestry.apache.org >>>> For additional commands, e-mail: dev-h...@tapestry.apache.org >>>> >>>> >>> >>> >>> >>> -- >>> Howard M. Lewis Ship >>> >>> Creator of Apache Tapestry >>> >>> The source for Tapestry training, mentoring and support. Contact me to >>> learn how I can get you up and productive in Tapestry fast! >>> >>> (971) 678-5210 >>> http://howardlewisship.com >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: dev-unsubscr...@tapestry.apache.org >>> For additional commands, e-mail: dev-h...@tapestry.apache.org >>> >>> >> >> >> -- >> Best regards, >> >> Igor Drobiazko >> http://tapestry5.de >> > > > > -- > Howard M. Lewis Ship > > Creator of Apache Tapestry > > The source for Tapestry training, mentoring and support. Contact me to > learn how I can get you up and productive in Tapestry fast! > > (971) 678-5210 > http://howardlewisship.com > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@tapestry.apache.org > For additional commands, e-mail: dev-h...@tapestry.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tapestry.apache.org For additional commands, e-mail: dev-h...@tapestry.apache.org