Estimating the amount of change between Pharo 2.0 and 3.0
Here is a simple calculation (that probably can be improved by the reflection or Moose aficionados) to estimate the amount of change between Pharo 2.0 and 3.0 by counting the number of methods in the image that have a timestamp later than 2.0's release date: these should be new or modified. (Object allSubclasses collect: [ :each | each methods size ]) sum. (Object allSubclasses collect: [ :class | ((class methods) select: [ :each | [ '2013-03-18T00:00:00' asDateAndTime <= (DateAndTime fromMethodTimeStamp: each timeStamp) ] on: Error do: [ false ] ]) size ]) sum. 14735/(71973/100.0) => ~20 % Not bad. Sven
2014-02-19 19:30 GMT-03:00 Sven Van Caekenberghe <sven@stfx.eu>:
Here is a simple calculation (that probably can be improved by the reflection or Moose aficionados) to estimate the amount of change between Pharo 2.0 and 3.0 by counting the number of methods in the image that have a timestamp later than 2.0's release date: these should be new or modified.
(snip)
14735/(71973/100.0) => ~20 % Not bad.
20% of change in Quantity terms is a lot! We'll never know what that 20% means in Quality terms. More than 20% for sure. Regards, -- E.
On Thu, Feb 20, 2014 at 2:49 AM, Esteban A. Maringolo <emaringolo@gmail.com> wrote:
20% of change in Quantity terms is a lot!
that's really a lot and is a bit frightening. -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." Winston Churchill
Why? The tests are green. When you change a complete compiler and AST then it is a lot of changes. Then each time you changes a comment you change the timestamp soâ¦.. Now 30 is stable since months. Stef PS: you know what is frightening is when you have a bug that arrives only after 2 weeks on heavy charged servers. We got that in the past and the changes producing it when just a couple of lines. So size is not a good predictor.
Pharo4Stef wrote:
Why? The tests are green. When you change a complete compiler and AST then it is a lot of changes. Then each time you changes a comment you change the timestamp soâ¦.. Now 30 is stable since months.
Stef
PS: you know what is frightening is when you have a bug that arrives only after 2 weeks on heavy charged servers. We got that in the past and the changes producing it when just a couple of lines. So size is not a good predictor.
What is frightening is the _ability_ of you guys to keep the system running each day as you make that amount of change :)
participants (5)
-
btc＠openinworld.com -
Damien Cassou -
Esteban A. Maringolo -
Pharo4Stef -
Sven Van Caekenberghe