On Wed, Sep 4, 2019 at 4:51 PM Xeno Amess <xenoam...@gmail.com> wrote:

> Well, if you do think repeating the similar codes for 100+ times is
> better than reflection and be more elegant or easier to read or
> something, then you can try maintain them.
> There is 100+ Throwable classes who have string constructor in JDK IMO.
> And don't forget some of them might only be in some certain versions
> of JDK, and be careful about making them compilable on other version
> of JDKs.
> Also you might try to use some preprocess way to generate such
> classes, some ways like lombok do (although that might be
> controversial).
>

Allow me to repeat, perhaps rephrase, and elucidate any miscommunication
based on what I said in the PR here:
https://github.com/apache/commons-lang/pull/450#issuecomment-527892606

I am _not_ suggesting to apply this pattern to 100+ exception classes,
ever. This is really an ever stricter application of the YAGNI and 80/20
rules. Both this idea and Commons Lang do not intend to provide this kind
of comprehensive coverage. This PR is for one exception and eats its own
dog food by refactoring 52 existing call sites within Commons Lang to use
the new methods. I then suggested the same could applied for
NullPointerException (7 possible call sites in Lang.) and
IllegalStateException (29 possible call sites in Lang.) but I am not doing
as far as implementing this, this could be done sooner or later. I really
think that by covering these three cases, we would fall into the 80/20
rule. In addition, there is reason to cover other exceptions like JDBC's
SQLException family of classes, but that's a job for Commons DbUtils or
Commons DBCP, not Commons Lang.

Gary


> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
> For additional commands, e-mail: dev-h...@commons.apache.org
>
>

Reply via email to