commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <>
Subject Re: [CSV] CSVFormat validation - do this earlier?
Date Wed, 14 May 2014 19:19:06 GMT
On 13 May 2014 16:53, Benedikt Ritter <> wrote:
> Hi,
> 2014-05-13 12:06 GMT+02:00 sebb <>:
>> At present, validate() is not invoked until the format is used. This
>> means that invalid arguments are not detected at the point they are
>> provided.
>> It would be possible to run validate as part of the withArgument()
>> methods. This would allow earlier detection, and avoid the issue that
>> currently some incorrect arguments may generate ISE rather than IAE.
>> I think the only possible user problem is that escape must be set
>> before using Quote.NONE as the policy, but this could be documented.
>> Otherwise AFAICT the arguments can be set in any order.
> Validating the format when it is constructed rather then when it is used is
> a good idea. But I think the only way to do this in a usable way is to use
> the builder pattern. There are several field that depend on each other
> during the validation. We don't want end up having user to look up the
> validate code in order to be able to create formats...

I agree it would be better to use the builder pattern, but I lost that argument.

I think it should still be possible to validate each argument
separately, even though they do depend on each other.

However, it would prevent some subsequent changes to formats unless
they were done in the correct order.
e.g. if one wanted to swap the escape and delimiter characters that
would be a bit tricky.
But I don't see the need to support all possible sequences of updates
to formats, so long as a format can easily be fully configured

> Benedikt
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail:
>> For additional commands, e-mail:
> --

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message