On 2016-10-09, Scott Kostyshak wrote:

> [-- Type: text/plain, Encoding:  --]

> On Sun, Oct 09, 2016 at 04:50:35PM +0000, Guenter Milde wrote:

>> * Changes must not only tested for compilation success but also for correct
>>   output.

> I don't think I would want to spend the time to manually inspect the
> output of all of the exports.

There is no need to test *all* exports, just the default (i.e. pdf2) and
only for changes that may change the output.

>> That said, I am not against the proposed changes (but Uwe is the one to
>> decide).

> Because of what I said above, I think you would be against the proposed
> changes. Correct?

> Perhaps there is some automatic way to do it, using a tool that looks
> for diffs in PDFs. diffpdf is useful, for example.

If you understand the consequences of a change, manual testing may be
confined to one (or none for simple changes) document (not all
translations, say).


Reply via email to