[tex-live] CTAN packages and generated files

[Joseph Wright]

On 02/09/2018 16:36, Manfred Lotz wrote:
> On Sun, 2 Sep 2018 15:49:57 +0100
> Joseph Wright <joseph.wright at morningstar2.co.uk> wrote:
> 
>> On 02/09/2018 15:46, Manfred Lotz wrote:
>>> Hi folks,
>>> For packages uploaded to CTAN we have a policy that in a package we
>>> don't like to have files which are generated by .ins or .dtx
>>> because we want to avoid redundancy.
>>>
>>> So we check for either   \generate{  \file{...}  ... } or
>>> generateFile{..} to find those files we then would complain about.
>>> Hope this is technically correct.
>>>
>>> A notable exception is the case where the mandatory README file
>>> (which actaually may be one of README, README.txt or README.md) is
>>> generated by a .dtx or .ins.
>>>
>>>
>>> Now, my question:
>>>
>>> It also could happen that an x.dtx contains a generate statement
>>> which mentions x.ins. Perhaps, also the other way around. Dunno.
>>>
>>>
>>> So, my thinking was if in x.dtx I find something like
>>>
>>> \generate
>>>     {
>>>       ...
>>>       \file{x.ins}
>>>       ...
>>>     }
>>>
>>> I just ignore that because x.dtx and x.ins are always required to
>>> exist as a pair. Otherwise LaTeX would complain when processing
>>> x.ins/x.dtx. Actually, this is the way checking takes place
>>> currently.
>>>
>>>
>>> Is this correct?
>>
>> No. It's possible to make a 'self-contained' .dtx file which has the
>> .ins instruction in the .dtx and generates everything in a single
>> "tex foo.dtx" run.
>>
> 
> Does it mean that in all cases where I find a .ins generate instruction
> in a .dtx file I may conclude that the .dtx is a 'self-contained' .dtx
> file?

I'd say so, yes.

Joseph