The QA Memory idea is interesting. I wonder how you distinguish “this area previously failed” from “the old failure is still relevant to the current architecture.” Otherwise historical evidence can become another form of stale context.
That's exactly what are the risks I'm definitely trying to avoid the idea isn't that like historical bugs permanently increase the risk in the same area or that QA memory stores just something more than this file failed before when it comes to changes the goal is to match against the context rather than the file history which basically helps in terms of when the architecture has changed enough would causes and requirements no longer applies the memory should lose relevance rather than keep influencing the verification and quality assurance forever the distinction between the useful historical context and stale context is definitely something I want to get right.
The QA Memory idea is interesting. I wonder how you distinguish “this area previously failed” from “the old failure is still relevant to the current architecture.” Otherwise historical evidence can become another form of stale context.
That's exactly what are the risks I'm definitely trying to avoid the idea isn't that like historical bugs permanently increase the risk in the same area or that QA memory stores just something more than this file failed before when it comes to changes the goal is to match against the context rather than the file history which basically helps in terms of when the architecture has changed enough would causes and requirements no longer applies the memory should lose relevance rather than keep influencing the verification and quality assurance forever the distinction between the useful historical context and stale context is definitely something I want to get right.