We should also unify it with --watch mode, so we will just have single auto-restart engine, but different reasons for auto-restart based on previous logic, this change is only refactoring, and it should keep all the code backward compatible if possible.
We also should double check that we have configuration of a way of auto-restart - via fresh context - or with --resume of the same context.
We also should add statistical information of total tokens used in saved conversation (based on actual json file), so later we will be able to decide smartly should we resume the same session or start anew.
I think that this is actually all --watch mode.
And we should add --watch-mode, which will have two values: restart and resume, and default. So user can experiment with different modes. Our refractoring also should include unification of all comments in watch mode, for example if mode is restart the comment title should be Auto-restart, and if mode is resume, the comment title should be Auto-resume.
default should be used if we have different configurations by default for auto-restart for until mergable and for auto-restart for uncommitted changes. We also should add new --watch-reasons, to allow user to set an lino notation style array of all modes we have, so the default value should be --watch-reasons 'uncommitted-changes until-mergable', and so on, may be in code we now have more experimental modes.
We need to unify all the logic under --watch mode, so later we will be able to support more reasons like new-issue-comment, new-pull-request-comment, issue-description-changed, pull-request-description-changed and so on.
Please collect data related about the issue to this repository, make sure we compile that data to ./docs/case-studies/issue-{id} folder, and use it to do deep case study analysis (also make sure to search online for additional facts and data), and propose detailed solution plan.
We should also unify it with
--watchmode, so we will just have single auto-restart engine, but different reasons for auto-restart based on previous logic, this change is only refactoring, and it should keep all the code backward compatible if possible.We also should double check that we have configuration of a way of auto-restart - via fresh context - or with --resume of the same context.
We also should add statistical information of total tokens used in saved conversation (based on actual json file), so later we will be able to decide smartly should we resume the same session or start anew.
I think that this is actually all
--watchmode.And we should add
--watch-mode, which will have two values:restartandresume, anddefault. So user can experiment with different modes. Our refractoring also should include unification of all comments in watch mode, for example if mode isrestartthe comment title should beAuto-restart, and if mode isresume, the comment title should beAuto-resume.defaultshould be used if we have different configurations by default forauto-restart for until mergableand forauto-restart for uncommitted changes. We also should add new--watch-reasons, to allow user to set an lino notation style array of all modes we have, so the default value should be--watch-reasons 'uncommitted-changes until-mergable', and so on, may be in code we now have more experimental modes.We need to unify all the logic under
--watchmode, so later we will be able to support more reasons likenew-issue-comment,new-pull-request-comment,issue-description-changed,pull-request-description-changedand so on.Please collect data related about the issue to this repository, make sure we compile that data to
./docs/case-studies/issue-{id}folder, and use it to do deep case study analysis (also make sure to search online for additional facts and data), and propose detailed solution plan.