Re: Java 1.4

-----Original Message-----
From: pgsql-jdbc-owner(at)postgresql(dot)org [mailto:pgsql-jdbc-owner(at)postgresql(dot)org] On Behalf Of Lew
Sent: Monday, January 23, 2012 1:21 PM
To: pgsql-jdbc(at)postgresql(dot)org
Subject: Re: [JDBC] Java 1.4

On 01/23/2012 12:34 AM, John Lister wrote:
> On 22 Jan 2011 Lew<noone(at)lewscanon(dot)com> wrote:
>
>>On 01/22/2012 02:11 PM, Oliver Jowett wrote:
>>>On 23 January 2012 07:59, Lew<noone(at)lewscanon(dot)com> wrote:
>>>>"The" vendor? There are more than one.
>>>Which other Java vendors do you think we should consider here?
>
>>All of them.
>
>>IBM markets several Java implementations (PC, Z/OS, ...). HP has at least one.
>>Oracle itself has several.
>
> I think the issue is particular JDK revisions. I would imagine IBM, et al stick to the jdk specs (or should be) and so long as we don't use any vendor specific extensions then it should work across the board. Even if it doesn't I imagine many of the other versions are commercial offering for which I don't have the money to purchase a licence and I'm sure others are in the same boat.

The issue to which I was responding was whether Postgres JDBC drivers should support Java 1.4. How does your comment impinge on that?

My reasoning is that if a substantial body of customers uses Java 1.4, and there is some evidence that this is true, then Postgres JDBC drivers should be available for it.

> I understand that large enterprise setups are where 1.4 is being used which is why I posed the question, but I would echo the sentiments of Till in an earlier post, that the people with a requirement for 1.4 are unlikely to use a new version of the driver without substantial testing or a new version of the database, and if they do are likely to have paid support... but I maybe wrong?

And my answer is yes, we should support Java 1.4, for the reasons I stated and to which you have not spoken.

-----------------------------------------------------------------------------

I'm only minimally following this but like most things specifics really do help. Is there a listing of not-yet-implemented features (or improvements)? Regardless, such a list should have minimum JDK release requirements listed so that the community would know which desirable features require either a discontinuance of support for older Java releases OR maintaining additional backward-compatible releases. There doesn't seem to be that many changes to the codebase that would overwhelm multiple distributions even if there is a smaller number of maintainers using a supposedly "not as branching friendly" source control infrastructure.

Skipping some of the language conveniences (i.e., generics) available in more recent versions is maybe annoying but it is a reasonable trade-off for maintaining backward compatibility; but if there is a unavoidable version restriction for a specific feature then it is likely to be better to implement that feature and up the minimum version than to ignore it in favor of either simplicity or keeping the newest release Java 1.4 compatible.

The same kind of thinking goes toward officially supporting different JREs. Make efforts to maintain broad support - though that technically means having these releases in the buildfarm - and address the pros/cons of specific features that are desirable but that cannot be used on all flavors.

My 2 cents.

David J.