Add some enum utilities #8171
Add some enum utilities #8171
Conversation
| /// the total number of items which have no matching enum key (holes in the enum | ||
| /// numbering). So for example, if an enum has values 1, 2, 5, and 6, max_unused_slots | ||
| /// must be at least 3, to allow unused slots 0, 3, and 4. | ||
| fn directEnumArrayLen(comptime E: type, comptime max_unused_slots: comptime_int) comptime_int { |
There was a problem hiding this comment.
This should also be offset by a min
There was a problem hiding this comment.
The intent of this api is to be directly indexed by the integer value of the enum, hence the direct part of the name. The max_unused_slots parameter ensures you don't have more padding than you intended. If you want a better packing, the EnumArray type will offset by a min.
| if (unused_slots > max_unused_slots) { | ||
| const unused_str = std.fmt.comptimePrint("{d}", .{unused_slots}); | ||
| const allowed_str = std.fmt.comptimePrint("{d}", .{max_unused_slots}); | ||
| @compileError("Cannot create a direct enum array for "++@typeName(E)++". It would have "++unused_str++" unused slots, but only "++allowed_str++" are allowed."); |
There was a problem hiding this comment.
This should probably return an error rather than a @compileError, otherwise a user cannot provide a fallback as an enum grows.
There was a problem hiding this comment.
I don't understand the use case for catching this error. This is a safety check to make sure that you are not accidentally making an array with a large number of empty slots. If this triggers, the way to handle it is to inspect it and decide whether to increase the allowed number of unused slots or switch to a denser packing like EnumArray. If you are writing generic code that calls this function, your code should accept this parameter alongside the enum.
Can you provide a more specific example where you would want to catch this?
daurnimator
added
the
standard library
|
Is this CI failure related to my changes? I don't see an error message so I'm not sure what the problem is. Maybe the additional comptime execution for the tests causing OOM? |
|
You can close and re-open the pr to re-run the ci. |
Absolutely no. |
I think that's it, unfortunately. master branch right now is taking a whopping 8.2 GiB to run std lib tests. This branch bumps it up to 8.4, which is hardly the fault of this code, but it seems to be the straw that broke the camel's back. I want to merge this though, so it seems I will have to find some other way to recover the extra memory. |
This PR adds some enum utilities, including EnumSet, EnumMap, and EnumArray data structures, and some helper functions that would have been useful in some recent enum-heavy work I have been doing. It conflicts with #7649 and #7882, which add similar functionality. However, this implementation is more complete and handles extern enums with multiple fields mapped to the same value.
I'm open to any suggestions for improvements or name changes.