> Also note that the Mira architects were actually looking at Jenkins at > that time, and pointed out that it was a bit old-fashioned in focusing > on "latest" versions by default and lacking proper changeset identification.
The story behind is actually more delicate: mira was started with the
following intentions:
a) provide a replacement for the isatest infrastructure with its
uncertain future at that time;
b) separate the »general idea« from the concrete application;
c) explore the design space for scheduling tests covering and combining
multiple histories.
The incentive to pursue c) led us to ignore Hudson/Jenkins entirely,
with the attitude »keep it simple, stupid«.
The following insights contributed to abandon the project:
1) Over time it became manifest that c) is too ambitious wrt. real
existing resources, hence the insight that pragmatically »tip testing«
is not so bad at all.
2) Further, »keep it simple, stupid« came to its limits with all the
involved technicalities: issues jobs on remote machines, exception
handling, proper daemonizing – it is surely easy to extend that list.
The Jenkins universe, as fas as I can tell, embodies significant
expertise in that area which you don't want to reimplement.
Cheers,
Florian
--
PGP available:
http://isabelle.in.tum.de/~haftmann/pgp/florian_haftmann_at_informatik_tu_muenchen_de
signature.asc
Description: OpenPGP digital signature
_______________________________________________ isabelle-dev mailing list isabelle-...@in.tum.de https://mailmanbroy.informatik.tu-muenchen.de/mailman/listinfo/isabelle-dev
