What are the main issue we are trying to solve with sample folder
structure ? I believe that, by having more granular distributions, the
amount of samples in a given distribution will be very small, and very
aligned with the purpose of the distribution.

Also note that in the past, our svn was structured as a folder
hierarchy which cause confusion to users and new-developers, hence we
moved to a flat module structure.

Just my $0.02

On Fri, Jan 23, 2009 at 3:52 AM, Simon Laws <[email protected]> wrote:
>
>
> On Fri, Jan 23, 2009 at 11:28 AM, ant elder <[email protected]> wrote:
>>
>>
>> On Fri, Jan 23, 2009 at 10:17 AM, Simon Laws <[email protected]>
>> wrote:
>>>
>>>
>>> On Fri, Jan 23, 2009 at 8:40 AM, ant elder <[email protected]> wrote:
>>>>
>>>>
>>>> On Thu, Jan 22, 2009 at 8:53 AM, Simon Laws <[email protected]>
>>>> wrote:
>>>>>
>>>>>
>>>>> On Thu, Jan 22, 2009 at 8:43 AM, ant elder <[email protected]> wrote:
>>>>>>
>>>>>> Related to the Samples thread [1] which is about how the samples work
>>>>>> internally we can also consider how they get included in a distribution 
>>>>>> and
>>>>>> thats come up in the distribution thread [2]. The main points there 
>>>>>> would be
>>>>>> that the samples are changed from all being in a single flat samples 
>>>>>> folder
>>>>>> to be grouped into sub folders, and that the sample names are rearranged 
>>>>>> to
>>>>>> point out their main reason for being.
>>>>>>
>>>>>> Adding some structure seems to me like a huge improvement to the
>>>>>> existing layout, there are so many samples now its become too hard and
>>>>>> confusing to see whats what. However this has come up before and I'm not
>>>>>> sure everyone agrees so before we spend much time does anyone want to 
>>>>>> raise
>>>>>> any concerns about a change like this?
>>>>>>
>>>>>>    ...ant
>>>>>>
>>>>>> [1] http://apache.markmail.org/message/yofc6tptydeszty6
>>>>>> [2] http://apache.markmail.org/message/uzmcywrpjctnlffy
>>>>>
>>>>> We'll I'm for It but I should point out I haven't done anything other
>>>>> than create some new samples with new names and updated the distro config 
>>>>> to
>>>>> group samples to get a feel for how it looks. The samples folder itself is
>>>>> still flat and IMHO I can probably live with that as long as we take time 
>>>>> to
>>>>> review the naming.
>>>>>
>>>>> Simon
>>>>
>>>> I'm not sure what you mean, you say you're all for it (Adding some
>>>> structure) but also that the samples folder itself is still flat? How would
>>>> that work?
>>>>
>>>>    ...ant
>>>>
>>>
>>> Locally I have a flat samples folder from svn and a structured samples
>>> folder in the distribution. Just configuration in the distro build makes
>>> this work.
>>
>> What are the advantages of doing it that way? If we agree some structure
>> in the distribution is useful it will be simpler to have svn match that wont
>> it?
>>
>>    ...ant
>>
>>
>
> The immediate advantage was that I could experiment without reorganizing
> svn. We haven't agreed that things should be grouped or if they are how that
> grouping should occur . I'm experimenting in order to inform my opinion.
>
> From our previous discussion my current position is that I'd like to have
> samples focus on demonstring the behaviour of a particular extension and
> hence my previous comments about sample naming. Doesn't mean that samples
> won't use many SCA extensions just that the focus would be on one particular
> feature, for example,
>
> implementation-java-calculator
> binding-ws-calulator
> host-webapp-calculator.
>
> Both binding-ws-calculator and implementation-webapp-calculator use
> implementation.java of course but that's not the focus of the sample or the
> documentation that supports the sample.
>
> Leading on from this I was starting to group samples as I mentioned
> previously and, with my experiment being based on where the distro stands at
> the moment + the addition of a webapp group, it happened to look like this
> this time round;
>
> core
>    implementation-java-calculator
> webservices
>    binding-ws-calculator
> webapp
>    host-webbapp-calculator
>
> However based on our discussion on the distribution thread I've dropped this
> actual structuring as that's not particularly what I'm interested in just
> now.
>
> I'm actually trying to get the ant scripts to run more automatically so the
> reason I came up with new tests names was so that I had some new tests where
> I could play with the ant script, pom files etc. I haven't posted anything
> yet but I will do shortly on a separate sample testing  thread.
>
> All this is just me thinkng aloud and is all up for discussion. I haven't
> checked any changes in. I will probably check in my experimental samples so
> I don't loose them.
>
> Simon
>



-- 
Luciano Resende
Apache Tuscany, Apache PhotArk
http://people.apache.org/~lresende
http://lresende.blogspot.com/

Reply via email to