I'd tend to agree, though caching in some situations has been the
only thing to save us from some serious API shortcomings. For
example, the fact that you cannot access the MavenProject instance
corresponding to a resolved Artifact...the cache in the project
builder is the only thing keeping Maven from having to re-parse that
POM (when you're trying to access the project instances for all
dependencies, this is non-trivial).
Piotr, I'd like to know more about the test you used to reach that
conclusion. Two seconds on a single (or two-module) build is
significant, especially when you consider that some people are
building over 100 modules at a time. If it's correct but dog-slow,
that's not really an acceptable solution, either.
-john
On Jul 12, 2007, at 2:36 PM, Carlos Sanchez wrote:
I have seen other caching problems when using the embedder, like
plugin versions not updated if the pom changes. I would go for
correctness first and speed second
On 7/12/07, John Casey <[EMAIL PROTECTED]> wrote:
So, if we don't ever cache the plugin container, it's a 2-second
penalty on a single project using debug output with corrected
algorithm?
-john
On Jul 12, 2007, at 11:29 AM, Piotr Tabor wrote:
> John Casey pisze:
>> While #3 might have the best performance in terms of speed, I
really
>> wonder about memory performance. Leaving all those old antrun
plugin
>> instances laying around but still reachable means that in very
large
>> projects that use antrun, we'll have some pretty significant
memory
>> problems. If you determine that the plugin+dependencies doesn't
fit
>> the current requirement, then what's the point in keeping it
around?
> We can use LRU cache if we afraid of memory.
>
> The running times are:
> 12 secunds with -X option for the "integration-test" with
> current "error"
> 15 secunds with -X option for the removing not matching
> (pluginDescriptor) and putting new one into the cache
> 17 secunds with -X option without cashing commented
>
> Thanks,
> Piotr
>
>
---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
---
John Casey
Committer and PMC Member, Apache Maven
mail: jdcasey at commonjava dot org
blog: http://www.ejlife.net/blogs/john
--
I could give you my word as a Spaniard.
No good. I've known too many Spaniards.
-- The Princess Bride
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---
John Casey
Committer and PMC Member, Apache Maven
mail: jdcasey at commonjava dot org
blog: http://www.ejlife.net/blogs/john
