Re: retrograding with convert-ly

[David Kastrup]

-Eluze <address@hidden> writes:

> dak-3 wrote:
>> 
>> Graham Percival <address@hidden> writes:
>> 
>>> On Mon, Jul 16, 2012 at 02:11:29AM +0200, David Kastrup wrote:
>>>> -d means no update in version header unless changes happen.  That is
>>>> also usually what you would want.  Without -d, the version of the last
>>>> applicable rule is used instead (rather than the last rule actually
>>>> causing a change).
>>>> 
>>>> In the case that no rule would be applied because the file is already
>>>> newer than all rules, I think it would make sense _not_ to change the
>>>> version header even without -d.
>>>
>>> If we did that, then people would complain "I'm using 2.16.2 but
>>> convert-ly only updates my file to 2.16.0!".
>> 
>> Which is exactly what is happening when the last rule of convert-ly is
>> for 2.16.0 while the current version is 2.16.2.
>> 
>>> This could be avoided by printing a message to the effect of "no
>>> changes to apply; not changing version number in the file".
>> 
>> Eluze converted for 2.15.41 with convert-ly from 2.15.41 and the file
>> already being at 2.15.41, and he complained that the version header was
>> set back to 2.15.40.
>> 
>>> As a general rule, I don't think it matters whether we make -d or not
>>> -d the default;
>> 
>> That was not even the question.
>> 
>>> what matters most is providing good information to the user in some
>>> combination of program output and/or documentation.
>> 
>> I prefer changing useless behavior over documenting it.
>> 
>> 
>
> +1
>
> first of all I'm relieved it is only a shallow bug and the functionality of
> convert-ly is not touched.
>
> the actual behavior is somewhat confusing, but tbh I only detected this
> after using it for several years.

It's somewhere after line 259 in scripts/convert-ly.py

-- 
David Kastrup