I finally worked out how to make my go-diff pkg <https://pkg.go.dev/github.com/mark-summerfield/go-diff> (v1.1.0) able to diff sequences of structs based on a key function.
On Friday, July 14, 2023 at 12:27:46 PM UTC+1 Mark wrote: > Hi Brian, > Ah, thank you, that helped a lot. > > On Friday, July 14, 2023 at 11:54:51 AM UTC+1 Brian Candler wrote: > >> The 'diff' package you showed identifies the changed object index by >> means of a "Path" attribute in the Changelog entry. If the top level object >> is a slice, then Atoi(change.Path[0]) is the index into the slice. >> >> On Friday, 14 July 2023 at 10:47:47 UTC+1 Mark wrote: >> >>> Hi Brian, >>> Your code certainly identifies the different items. >>> However, that's not a diff tool in the sense I mean. >>> Unix diff and tools like it don't just say x[i] != y[i], they find the >>> longest common subsequences and in essence produce a series of edit >>> commands that would turn slice x into slice y. >>> There are quite a few go diff tools that will do this, including my own >>> <https://pkg.go.dev/github.com/mark-summerfield/go-diff> based on >>> Python's difflib sequence matcher. >>> What I want to do is find one that does this for slices of structs where >>> only one struct field is considered for comparison purposes. >>> >>> On Friday, July 14, 2023 at 10:00:34 AM UTC+1 Brian Candler wrote: >>> >>>> I forgot you wanted generics: >>>> https://go.dev/play/p/PhGVjsWWTdB >>>> >>>> On Friday, 14 July 2023 at 09:47:21 UTC+1 Brian Candler wrote: >>>> >>>>> You seem to be saying "if the S field is different then I want to >>>>> consider these two structs different, and get pointers to the two >>>>> structs. >>>>> If the S field is the same then I want to skip the pair entirely". Is >>>>> that >>>>> right? >>>>> >>>>> The required semantics are not entirely clear, but it sounds like a >>>>> handful of lines of code to implement - there's no point importing and >>>>> learning a third party library. >>>>> >>>>> On the assumption that all the elements to be compared are in >>>>> corresponding positions in a and b: >>>>> https://go.dev/play/p/Y71sLUpftzR >>>>> >>>>> On Friday, 14 July 2023 at 09:11:35 UTC+1 Mark wrote: >>>>> >>>>>> In fact the diff pkg mentioned above does work but is of no use to me >>>>>> since for each change it gives back only the field(s) used, not the >>>>>> original structs (or pointers to them), so I can't see any way back to >>>>>> the >>>>>> original structs (or their slice indexes). >>>>>> >>>>>> On Friday, July 14, 2023 at 8:58:41 AM UTC+1 Mark wrote: >>>>>> >>>>>>> What I really want to do is to be able to diff slices of structs on >>>>>>> the basis of one single field. >>>>>>> For example, given: >>>>>>> ``` >>>>>>> type Item struct { >>>>>>> I int >>>>>>> S string >>>>>>> } >>>>>>> ``` >>>>>>> and given `a` and `b` are both of type`[]Item`, I want to diff these >>>>>>> slices based purely on the `S` field, ignoring the `I` field. >>>>>>> >>>>>>> This diff pkg <https://pkg.go.dev/github.com/r3labs/diff/v3> claims >>>>>>> to be able to do this (something I'm testing, so I don't know either >>>>>>> way >>>>>>> yet), but in any case, it is incredibly slow. >>>>>>> >>>>>>> On Friday, July 14, 2023 at 8:31:39 AM UTC+1 Peter Galbavy wrote: >>>>>>> >>>>>>>> As a slight digression - I thought I was going mad, but 'slices' >>>>>>>> and 'maps' are new :-) Only in 1.21 though... >>>>>>>> >>>>>>>> Well, there is a lot of boiler plate that maps.Keys() will get rid >>>>>>>> of. >>>>>>>> >>>>>>>> On Thursday, 13 July 2023 at 10:06:01 UTC+1 Brian Candler wrote: >>>>>>>> >>>>>>>>> Structs are already comparable, but all fields must be the same: >>>>>>>>> https://go.dev/play/p/XwhSz4DEDwL >>>>>>>>> >>>>>>>>> I think your solution with function 'eq' is fine. You can see the >>>>>>>>> same thing in the standard library in slices.CompactFunc and >>>>>>>>> slices.EqualFunc >>>>>>>>> https://pkg.go.dev/slices#CompactFunc >>>>>>>>> https://pkg.go.dev/slices#EqualFunc >>>>>>>>> >>>>>>>>> For the case of "ordered" rather than "comparable", have a look at >>>>>>>>> slices.BinarySearchFunc and related functions. >>>>>>>>> >>>>>>>>> On Thursday, 13 July 2023 at 09:29:38 UTC+1 Mark wrote: >>>>>>>>> >>>>>>>>>> I have a package which has a function `Do[T comparable](a, b []T) >>>>>>>>>> Result`. >>>>>>>>>> I have a struct: >>>>>>>>>> ```go >>>>>>>>>> type N struct { >>>>>>>>>> x int >>>>>>>>>> y int >>>>>>>>>> t string >>>>>>>>>> } >>>>>>>>>> ``` >>>>>>>>>> Is it possible to make `N` comparable; in particular by a field >>>>>>>>>> of my choice, e.g., `t`? >>>>>>>>>> >>>>>>>>>> Or will I have to make, say, `DoFunc(a, b []N, eq func(i, j N) >>>>>>>>>> bool) Result` with, say, >>>>>>>>>> `func eq(i, j N) { return i.t == j.t }`? >>>>>>>>>> >>>>>>>>> -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/873970ab-d4af-4cb3-bec1-ba91e5e64819n%40googlegroups.com.
