README: "Stability" section: Add types of errors
This commit is contained in:
parent
99eded8486
commit
a793bd609a
|
@ -73,7 +73,13 @@ The project is carefully developed, however, it's practically impossible to guar
|
||||||
1. automated testing is not feasible; although debootstrap installations could be automated, the bulk of the work is related to the installers, which can't be automated without sophisticated GUI automation,
|
1. automated testing is not feasible; although debootstrap installations could be automated, the bulk of the work is related to the installers, which can't be automated without sophisticated GUI automation,
|
||||||
1. testing is time consuming, so it can be performed on a limited amount of distros at a time.
|
1. testing is time consuming, so it can be performed on a limited amount of distros at a time.
|
||||||
|
|
||||||
Errors due to installer will cause the script to terminate, so, generally speaking, if the script completes, the system has been successfully setup.
|
Broadly speaking, there are two types of breakages:
|
||||||
|
|
||||||
|
1. minor changes directly or indirectly related to the installer, for example:
|
||||||
|
- partition mounts change behavior (e.g. when they're dismounted)
|
||||||
|
- installed services change behavior (e.g. a new service creates an ephemeral file under /target/run, and the sync fails because the file disappears)
|
||||||
|
1. GRUB setup not working
|
||||||
|
- most annoying issue to debug; the installer will succeed, but the installed O/S won't boot
|
||||||
|
|
||||||
## Demo
|
## Demo
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue