id_gen.go too large about opcua

[sruehl]

I named common.go to id_names_gen.go in the PR

[sruehl]

Edit: move PR comment on PR #680

[kung-foo]

I'm kinda on the fence about this. On one hand, I don't like adding complexity and code paths specifically to address IDE and web interface deficiencies, on the other hand, I also use Goland and have hit this issue before (though I "solved" it by bumping up Goland's limits). @magiconair thoughts?

[sruehl]

I can follow the argument and usually have the same stance that it doesn't make any sense to alter code because some IDE is unhappy. But even using VI or any editor this file is just too big 😄.
One note: I utilize the categories (type or whatever edit: NodeClass) the OPC-UA spec provides in the last column of the CSV. So from the spec those IDs are split up already. Splitting them up even more like I suggested would then fall again in the argument I brought up in the first sentence. So practically besides issues in Goland I had trouble generating code from it in plc4x using Java as it would either overflow the maximum number of enums supported 65535 or it would just overflow the static initalizer with too much bytes.
But as I wrote above: It doesn't hurt to segment. The imports/api look completely identical it is just split up into multiple files and makes it even more readable using any editor/viewer.

(edit: referring to #680 in this comment)

[magiconair]

Hmm, this works just fine in VSCode and neovim. Isn't this a limit that you can tune in Goland? Seems like an arbitrary limit.

I could live with splitting the enums and the map into two different files but making a random split because of a Goland limit seems pretty weird. Let's try to fix the IDE first.

[magiconair]

I also don't follow the code-generation limit argument. The number of enum values does not change. So why would splitting into multiple files not hit that limit?

Also, since the code generator is java you should be able to just provide more memory with -Xmx or not?

I don't think we had such a situation before and I'm also a bit on the fence here. On the one hand this code should just work. On the other hand I don't know how many people actually have this problem and whether the ones who do haven't just updated their settings and moved on. That's what I would do at least.

We can add something to the README if there is a simple workaround. If there isn't and the library is otherwise unusable with either tool then we should fix it.

[magiconair]

If we have to split it up then I'd suggest we do this by id since this doesn't require any complex semantics or regular expressions. For example

id_00000_gen.go
id_10000_gen.go
id_20000_gen.go
...
id_names_gen.go

[magiconair]

@sruehl is this still relevant?

[magiconair]

I've just asked a friend to check this and he says that there is a button to enable the syntax highlighting with the suggestions to bump the memory and the instructions to do so.

[sruehl]

yeah, but as I said this is still a pretty big file and splitting it up with the categories they supply upstream (#680) helps readability.
So even if I fix it for my particular IDE, Github still doesn't render it well.