[Oberon] A2 development and distribution; was, Distribution practice.

[peter at easthope.ca]

From:	Felix Friedrich <felixf at inf.ethz.ch>
Date:	Mon, 7 Nov 2022 08:22:19 +0100
> ... boils down to a binary search in past revisions until the 
> revision can be found that introduced the new behavior.

Fundamental methods are so pervasive and powerful!  A few days ago I 
considered binary search to find a fault in series wired Christmas 
lights.  In that case another algorithm was more appropriate but at 
the end of the day, 210 lights worked again.  =8~)

> Example: in revision x we change the representation of arrays with 
> pointers. Only in revision y = x + d we find out that on particular 
> systems ... We check out revision y - 200 ...

Difficulty of the correction process increases with d.  With a 
relatively low frequency of use, A2 has large values of d, in 
comparison to Debian Linux for example.  What strategy can help to 
keep d small for A2?  My naive thought is to increase rate of testing 
in relation to rate of revision.    Perhaps enlist volunteers to 
perform a test procedure for each platform.  I can test on Linux32 and 
according to 
https://en.wikibooks.org/wiki/Oberon/A2#Target_Systems_and_File_Name_Extensions
volunteers are needed for five other targets.  We also need a test 
process suitable for the volunteers.

Regards,                               ... P.L.


mobile: +1 778 951 5147
  VoIP: +1 604 670 0140
https://en.wikibooks.org/wiki/User:PeterEasthope