Higher fidelity imports #21114
Higher fidelity imports #21114
Conversation
techknowlogick
added
the
type/enhancement
GiteaBot
added
the
lgtm/need 2
GiteaBot
added
lgtm/need 1
| PosterID int64 `yaml:"poster_id"` | ||
| PosterName string `yaml:"poster_name"` | ||
| PosterEmail string `yaml:"poster_email"` | ||
| Created time.Time | ||
| Updated time.Time | ||
| Content string | ||
| Reactions []*Reaction | ||
| Meta map[string]interface{} `yaml:"meta,omitempty"` // see models/issues/comment.go for fields in Comment struct |
There was a problem hiding this comment.
How does the value of Meta gets updated? I can only see some code reading it, but I can not find code writing it.
If let me decide, I would use clear field names instead of the dynamic map: the field names are clearly known, the dynamic map make the code like a dynamic language which loses the static compiling time check
There was a problem hiding this comment.
It is written to via custom tooling to comments.yml and used during "restore-repo" CLI command.
I used an interface because this in theory could be extended to fit each field in the comments struct and I can import that struct due to cyclical imports.
There was a problem hiding this comment.
It is written to via custom tooling to comments.yml and used during "restore-repo" CLI command.
Is there a related PR to show how to use it?
I used an interface because this in theory could be extended to fit each field in the comments struct and I can import that struct due to cyclical imports.
I do not think the "in theory could be extended to fit each field" could really work with a dynamic map. In the end, on the reader side, every field is still pre-defined and clearly hard-coded in code. If it is the case, why not put some pointers like: AssigneeID *uint64, OldTitle *string.
|
closing in favour of #22510 |
Currently if you want to track the history of various things when importing, they'd be added in as comments (ie, x added y label to issue).
This allows for additional fields in yaml, specifically
issue comment type, as well as an interface rather than having to re-iterate all the ones that exist in the comment struct already.