commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Otto Fowler <ottobackwa...@gmail.com>
Subject Re: [DISCUSS] new component for timing?
Date Tue, 20 Mar 2018 21:30:35 GMT
I would love to get this on track.  I apologize if I have made it more
confusing than it needs to be,
I’m trying to be open to all the suggestions.

If we assume that stack watch is worth ‘having’, then the question is where
to put it.
commons-monitoring / sirota seems to me to be a ‘complete’ solution as
opposed to
a set of or collection of like classes.

Setting the community support / project aspect of sirota aside, it seems
strange to put
a separate class into a more complete and uniform system.  Unless there is
some generically
useful set of timing utility classes that could be taken out of sirota to
go into commons-????,
along with things identified ( StopWatch?) out of commons lang and possibly
other commons projects.

commons-timing seems reasonable.  Thoughts?



On March 17, 2018 at 11:24:32, Romain Manni-Bucau (rmannibucau@gmail.com)
wrote:

Yes but consequence was a lack of community increase which is a killer for
an incubator project on the long run.

Le 17 mars 2018 15:19, "Gilles" <gilles@harfang.homelinux.org> a écrit :

> On Sat, 17 Mar 2018 12:47:40 +0100, Romain Manni-Bucau wrote:
>
>> Le 17 mars 2018 11:49, "Gilles" <gilles@harfang.homelinux.org> a écrit :
>>
>> On Thu, 15 Mar 2018 15:13:39 +0100, Romain Manni-Bucau wrote:
>>
>> 2018-03-15 14:36 GMT+01:00 Otto Fowler <ottobackwards@gmail.com>:
>>>
>>> If we can come to consensus on the way forward, I’ll be happy to do the
>>>
>>>> work ( although I’ll need help of course ).
>>>> I guess I’m the straw that broke the camel’s back then? ;)
>>>>
>>>>
>>>>
>>>>
>>>> On March 15, 2018 at 08:09:45, Gilles (gilles@harfang.homelinux.org)
>>>> wrote:
>>>>
>>>> Hi.
>>>>
>>>> On Thu, 15 Mar 2018 03:52:58 -0700, Otto Fowler wrote:
>>>> > I think bringing back commons-monitoring/sirota would only be
>>>> > possible if
>>>> > it were to be modular enough that you could bring in the ‘core’
>>>> > classes
>>>> > without needing to bring in all of what sirota ended up being, which
>>>> > was an
>>>> > end to end solution.
>>>>
>>>> Isn't it possible? [I didn't look; Romain should tell whether he
>>>> would be interested in taking that route.]
>>>>
>>>>
>>>> Sirona was done modular, just the API, the in memory part, etc...
>>> But this kind of impl needs way more just after so not sure it does
worth
>>> splitting it to put it back altogether after.
>>>
>>> What is the rational to try to push a very small part @commons instead
of
>>> creating a community @incubator with an E2E solution? This is what I
fail
>>> to see.
>>> My experience - coming exactly from here - tends to make me think
commons
>>> will not fit very long or will not bring enough value pretty quickly
but
>>> that's just my opinion.
>>>
>>>
>> Not "just" an opinion since you evoke Sirona's precursor as being
>> the kind of component we'd reinstate. Unless we learn
>> 1. why the precursor needed to become TLP, and
>> 2. why the TLP didn't succeed,
>> we'd go in circles.
>>
>>
>> We failed at community@asf and pby communication/promotion levels I
>> think.
>> Other parts were successful (prod etc).
>>
>>
> [It seems that part of your message went missing.]
>
> Lack of marketing skills should not entail failure, especially
> not since communication is a transverse concern.
>
> Gilles
>
> Would it make sense that Sirona becomes (again?) "Commons Monitoring"?
>> Does the "StackWatck" (Otto's contribution that started this discussion)
>> already exist in a Sirona module? If not, can it be done, so that usage
>> is similar to what Otto had in mind?
>>
>> Regards,
>> Gilles
>>
>>
>> commons-monitoring or commons-timing seem to be the correct thing
>>>
>>>> > however,
>>>> > but I would like to think that there would be more impetus
>>>>
>>>> I'm afraid that it's rather the lack of manpower.
>>>> [And my inner conviction is that that state of things often
>>>> led to rush to cramming more code into existing components,
>>>> rather than "distribute" more uniformly according to subject
>>>> matters. When scarce human resources ("community") disappear,
>>>> cruft accumulates, sometimes up to stifle clean-up, maintenance,
>>>> improvement, and even development.]
>>>>
>>>> > to do this than
>>>> > thinking StackWatch is ‘too big’ for lang.time.
>>>>
>>>> It isn't any more than many other functionalities that were
>>>> introduced but shouldn't have been.
>>>> Depending on what the "Commons" PMC wants to favour ("code"
>>>> *or* "community"?), the choice is between continuing with the
>>>> accumulation, or back-pedaling through the creation of as
>>>> many *real* components as they are developers willing to
>>>> maintain them.
>>>>
>>>> > It really isn’t that complicated a thing.
>>>>
>>>> Sure.
>>>> The issue is somewhere else.
>>>> Note that, personally, I hadn't imagined that there would
>>>> be an issue for regular developers of [Lang] (or I wouldn't
>>>> have spent time reviewing the "details" ;-).
>>>> But I of course agree that the question should be asked; the
>>>> more so that, with [Math], we've a striking example of what
>>>> awaits a library that lacks boundary checks and explicit
>>>> road map.
>>>>
>>>> Regards,
>>>> Gilles
>>>>
>>>> > On March 8, 2018 at 11:50:17, Gilles (gilles@harfang.homelinux.org)
>>>> > wrote:
>>>> >
>>>> > On Thu, 08 Mar 2018 16:03:24 +0000, Gary Gregory wrote:
>>>> >> -1 to "commons-misc". It feels to me like a copout and unfocused
>>>> >> like
>>>> >> SomethingUtils.
>>>> >> We need a proper home.
>>>> >
>>>> > +1
>>>> >
>>>> >> How about the idea of commons-measure.
>>>> >
>>>> > Just because the first feature would happen to be a timer?
>>>> > What other content do you foresee?
>>>> >
>>>> >> Then there
>>>> >> still the idea of resurrecting other Apache projects. Kind of going
>>>> >> in
>>>> >> circles...
>>>> >
>>>> > Indeed, IIRC the questions were asked (whether the feature could
>>>> > be contributed to ex-Sirona and whether that project would be
>>>> > repatriated to "Commons") but not answered (unless I'm mistaken)...
>>>> >
>>>> > Best,
>>>> > Gilles
>>>> >
>>>> >
>>>> >> Gary
>>>> >>
>>>> >> On Mar 8, 2018 08:58, "Otto Fowler" <ottobackwards@gmail.com>
wrote:
>>>> >>
>>>> >> So, could think about commons-misc or something?
>>>> >> I don’t think we are going to come up with a perfect module for
>>>> >> these
>>>> >> things.
>>>> >>
>>>> >> Maybe the way it can work is:
>>>> >>
>>>> >> commons-misc exists.
>>>> >>
>>>> >> It is the landing place for things that seem to be outside the
scope
>>>> >> of
>>>> >> commons-xxxx, but don’t justify
>>>> >> a new module or sandbox effort.
>>>> >>
>>>> >> Things in misc can be reevaluated for inclusion in new modules at
>>>> >> things
>>>> >> go, and at that point @Depricated
>>>> >> out of misc.
>>>> >>
>>>> >> ?
>>>> >>
>>>> >>
>>>> >>
>>>> >> On March 3, 2018 at 00:42:12, Matt Sicker (boards@gmail.com) wrote:
>>>> >>
>>>> >> On 2 March 2018 at 13:31, Oliver Heger
>>>> >> <oliver.heger@oliver-heger.de>
>>>> >> wrote:
>>>> >>>
>>>> >>> One other suggestion: It was stated in the past that the
concurrent
>>>> >>> classes are also a bit out of scope for [lang], especially the
>>>> >>> circuit
>>>> >>> breaker implementations. Would it make sense to move those into
a
>>>> >>> new
>>>> >>> module, and could this be a home for the watch classes, too?
>>>> >>>
>>>> >>
>>>> >> Considering the amount of retry libraries there are out there, I
>>>> >> think it
>>>> >> makes perfect sense for circuit breaker libraries to be their own
>>>> >> thing,
>>>> >> too. See Hysterix for example.
>>>> >>
>>>> >> --
>>>> >> Matt Sicker <boards@gmail.com>
>>>>
>>>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
> For additional commands, e-mail: dev-help@commons.apache.org
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message