Another setup v2.889 regression?

[Jon Turney]

On 05/03/2018 12:50, Houder wrote:
> On 2018-03-04 23:25, Jon Turney wrote:
>> On 04/03/2018 15:11, Houder wrote:
>>> Because there are problems with setup v2.889, I decided to do
>>> some testing of my own.
>>
>> Thanks for reporting this.
>>
>>> 2018/03/04 15:35:20 Augmented Transaction List:
>>> 2018/03/04 15:35:20    0 install  cscope          15.8.0.1-2
>>> 2018/03/04 15:35:20    1   erase  cscope             15.8b-1
>>>
>>> Â From override.hint on the "Twente" mirror:
>>>
>>> curr: 15.8b-1
>>> prev: 15.8.0.1-2
>>> replace-versions: 15.8.0.1-2
>>>
>>> As far as understand, setup should attempt to install v15.8b1
>>> in case v15.8.0.1-2 is installed.
>>>
>>> Not the other way around. (currently v15.8b-1 in installed).
>>
>> Yeah, this isn't working as expected. Time for a bigger hammer...
>>
>> The previous version has been re-designated as 15.8a-2 (to match what
>> it identifies itself as with 'cscope -V').
>>
>> This should make setup update to the most recent version, while still
>> allowing the previous version to be offered as installable...
> 
> Uhm, bigger hammer ... you mean your modification will solve the problem
> for cscope, but not for f.i. the *lz4* packages?

Sigh.  I thought I had fixed lz4 [1], but apparently not, it suffers 
from the same class of problem (briefly, replace-versions: doesn't work 
on versions which are also offered for installation)

These are the only other packages which I am aware of which suffer from 
this specific problem, and I'm not sure I want to fix those in the same way.

If you think there are other package which have this problem can you 
please tell me what they are...

[1] https://cygwin.com/ml/cygwin/2018-02/msg00306.html

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple