Обсуждение: Release for 9.5

> I'd like to not use postgresql's version number in our artifacts.

+1

Vladimir

On Thu, Nov 12, 2015 at 1:55 PM, Vladimir Sitnikov
<sitnikov.vladimir@gmail.com> wrote:
>> I'd like to not use postgresql's version number in our artifacts.
>

I agree. The only issue is the numbering is used on some Maven
repository sites to indicate new versions. If we go to a version
number less than the current one, it might not be recognised.

It might be a good opportunity to adopt http://semver.org/ too?

> On Nov 12, 2015, at 2:59 PM, Dave Cramer <pg@fastcrypt.com> wrote:
>
>
> On 12 November 2015 at 17:35, Stephen Nelson <stephen@eccostudio.com> wrote:
> On Thu, Nov 12, 2015 at 1:55 PM, Vladimir Sitnikov
> <sitnikov.vladimir@gmail.com> wrote:
> >> I'd like to not use postgresql's version number in our artifacts.
> >
>
> I agree. The only issue is the numbering is used on some Maven
> repository sites to indicate new versions. If we go to a version
> number less than the current one, it might not be recognised.
>
> It might be a good opportunity to adopt http://semver.org/ too?
>
> fine with me.

+1 to semver, although it seems unlikely that we'd change the major version (backwards compatibility break)
anytime soon

>
> Does the version number need to be in the artifact name or just the version in the pom?

Nope, just the version -- the artifact name should stay constant

+1 to not having to match versions but we should have a version compatability matrix. As new features are added to core
PG,thereneeds to be a clear indication of where they're supported.  

As of now the recommendation is to always use the latest driver. As new data types of other features are supported, it
shouldbe clear which driver versions support which core DB features. Rather than doing it at a feature level I think it
makesmore sense to do it at a DB version level (since PG doesn't backpack new features). 

Similarly, if a legacy feature becomes unsupported, for example if we finally remove v2 protocol support, that should
beclearly marked.  

As far as actual version numbers go, if the maven coordinates are not going to change then we will need to ensure they
aresemver greater than the existing version. Otherwise it'll cause a world of comparability issues.  

Regards,
Sehrope Sarkuni

Dave Cramer wrote:

>  Does the version number need to be in the artifact name or just the version
>  in the pom?

It doesn't have to be there. Though maven includes the version number in the
artifact name by default. And I'd strongly advice against diverging from that.

If nothing else, having the version number in the artifact name makes it easy
to determine which version is used and that your app does not have inconsistent
transitive dependencies in a complex build environment.

--
Fredrik Jonson

> If nothing else, having the version number in the artifact name makes it easy
>to determine which version is used and that your app does not have inconsistent

Fredrik, you mean "build artifact should include version like
pgjdbc-10.20.30.jar", don't you?

>  Does the version number need to be in the artifact name or just the version
>  in the pom?

Dave, can you please clarify what do you mean by artifact?
Do you mean "the final jar file" or do you mean
<artifactId>...</artifactId> pom.xml tag?


Vladimir

Should we start from 42 then?

Vladimir

> the answer to everything

Vladimir

On 14/11/15 03:29, Vladimir Sitnikov wrote:
> Should we start from 42 then?
>
> Vladimir
>
>
This number is Sacred and should not be used in vain, lest Douglas Adams
comes back and haunts you!

It should be reserved for something really important, like a version of
PostgreSQL that is totally bug free, perfectly parallel in execution,
and lacks no features that sane DBA's would ever want - the last version
of PostgreSQL ever needed to be released: 12.3.42.

=Gavin

Vladimir Sitnikov wrote:

>  Fredrik, you mean "build artifact should include version like
>  pgjdbc-10.20.30.jar", don't you?

Yes, I ment the build artifact, i.e. the jar file. Not the artifactId attribute
in the pom.

Seems I misinterpreted Dave's question. And here I thought I was helping. <:)

--
Fredrik Jonson

> On Nov 13, 2015, at 1:51 PM, Gavin Flower <GavinFlower@archidevsys.co.nz> wrote:
>
>> On 14/11/15 03:29, Vladimir Sitnikov wrote:
>> Should we start from 42 then?
>>
>> Vladimir
> This number is Sacred and should not be used in vain, lest Douglas Adams comes back and haunts you!

Haha. Though we definitely can have a 42 in their to reference the latest version of the JDBC spec (4.2).

Regards,
Sehrope Sarkuni