Wow, that's not half as bad as I thought it would be. If you'd like to move
forward with contributing it, the next step is creating a JIRA ticket and
having someone (probably me) dig through it.

On Fri, Jul 23, 2010 at 6:38 AM, Emmanuel Bourg <[email protected]> wrote:

> Le 22/07/2010 22:35, Bryan Duxbury a écrit :
>
>
>  One question I have in general, though, is why does the size of the jar
>> matter at all? It's not like it's going to take up tons of memory or slow
>> you down particularly when being transferred around.
>>
>
> Thank you for the look at my suggestion.
>
> I wouldn't bother reducing the size of the jar for a server application,
> but for a desktop application I have to care about the size of the files
> downloaded. I guess 3MB for a RPC interface was beyond my threshold of pain.
>
> The performance is definitely important. In my opinion it's more important
> on a server handling several requests continuously than on a client
> launching some requests sporadically. It really depends on the use cases.
>
> I took some time to run my modification through the serializer benchmark
> [1]. Without optimization it's 46 times slower than the static code. With a
> naive cache it's only 2.3 times slower, which is acceptable for my use case.
> And there is certainly room for improvements with dynamic bytecode
> generation.
>
> Emmanuel Bourg
>
>
> [1] http://github.com/eishay/jvm-serializers
>
>

Reply via email to