We remove assisted inject and multi bindings and noticed improvements as well.
-- Brian
Sent from my iPhone
> On Oct 14, 2014, at 11:43 AM, Stephan Classen wrote:
>
> We used Guice3.0 in an library which analyzes SQLs sent to the DB. The first
> version showed a heavy impact on execution tim
Worth pointing out that in a cloud-universe where instances are killed and
restarted, improving startup time is actually money on the table, since
work performed by guice at startup costs CPU time which most cloud vendors
charge for. Perhaps a trickle, but still a "good thing" to fix where Guice
(o
We used Guice3.0 in an library which analyzes SQLs sent to the DB.
The first version showed a heavy impact on execution time (queries
were up to 4 times slower).
We replaced almost all assisted inject with hand written factories
and could remove about half of the overhead
Prime MVC calls getInstance every request. This is great because all
controllers are non-singletons and state doesn't get polluted. Though Guice is
rarely the bottleneck for most apps.
-- Brian
Sent from my iPhone
> On Oct 14, 2014, at 10:59 AM, Kevin Burton wrote:
>
> OK.. silly question,
Perf improvements for Guice tend to fall in two categories: Speed up
startup, and speed up runtime. Speeding up the startup makes the
development lifecycle better -- devs can start/stop/restart their servers
faster. Speeding up runtime has more obvious improvements: you can do more
things with f
OK.. silly question, why would this matter?
The performance of Guice didn't seem pathologically broken.
What I do is just create all my bindings at startup and then I'm done. Any
complex initialization is done via providers...
So maybe I'm missing something and there are some use cases tha
The user list is perfectly fine, the dev list isn't really actively used.
It's more just a place where commit messages go to.
sam
On Tue, Oct 14, 2014 at 10:05 AM, Stéphane NICOLAS
wrote:
> Thx for this answer Sam.
>
> I believe we will soon come back to guice-dev list with a suggestion of
> op
Thx for this answer Sam.
I believe we will soon come back to guice-dev list with a suggestion of
optimization for Guice. This week probably.
Stéphane
On Tuesday, October 14, 2014 9:52:48 AM UTC-4, Sam Berlin wrote:
>
> I don't think we have any exact measurements, but there were a number of
>
I don't think we have any exact measurements, but there were a number of
changes designed to improve memory & performance (that I believe were
individually measured). These are the ones I can spot by looking over the
commit history:
*
https://github.com/google/guice/commit/117b2a6bfe7fe6a59578e
Hi Guice team,
we have done some little experiments on our side and it seems Guice 4 is
slighlty faster than Guice 3. But you don't have much data about this.
Unfortunately, there is no release changes between Guice 3 and the upcoming
Guice 4 yet.
Does anyone have any measurements of performanc
10 matches
Mail list logo