Duplicate declaration 'xxxxxxx' in current scope

[jvaldiviezo9]

Describe the bug
Issue related to the new version 1.7.2
Getting an error "Duplicate declaration 'xxxxxxx' in current scope"

To Reproduce
Not sure how to reproduce, I'm working on a large project, however, this issue didn't happened on the past version 1.5.13

Code (if relevant)
-- Attached screenshot --

Expected behavior
I think the "Duplicate declaration" for an "Attribute VB_Name" item should not be checked for validation.

Screenshots

Additional context
I have a large project with multiple nested folders, at some point the error happened in other module named "main_timer". So I guess, somehow the parsing gets a conflict when trying to load variables. Possible related to the new "Definition Provider". Hope this helps.

[SSlinky]

Thanks for raising. It was definitely on my radar but I finally had some time to look at it properly today.

[SSlinky]

Let me know if it's still an issue in 1.7.4.

[jvaldiviezo9]

Hi @SSlinky,

Thanks, seems like the issue is partially solved in 1.7.4.
Let me explain:

Initially seems like all is okay, but I have some version control based on Git. If for example I revert back a file, like "Discard Changes", then it's like the overall project lost the previous status and the same error pop ups again. So, I guess that maybe the "intial scan" keeps linked to a "cache" that keeps storing the previous reference even after it's replacement with Git. So when the file is restored to a previous version by Git, it thinks that it's a different file and points it as a "Duplicate declaration".

In the same way, I'm noticing that to be able to use ctrl + click it takes some time. As I've mentioned in a different thread #48
I'm using Naive Definitions extension: https://marketplace.visualstudio.com/items?itemName=FallenMax.naive-definitions

Currently seems like the "Ctrl + Click" or "Go to Definition" feature doesn't work until the "initial scan" completes, which takes a lot of time (maybe like 2 minutes). Is there a workaround? This didn't happened on the earlier 1.5.13 version. I was able to immediately use that feature.

As an additional note, while testing, I've disabled the "Naive Definitions" extension, and the F12/Definition Provider seems not to be working yet (or at least in a large project).

Thanks again!

[SSlinky]

The initial scan reads the files from disk and interprets the AST into memory. It shouldn't be instant, but I am not expecting it to take two minutes. Could you raise a separate issue for that and let me know what repo you're scanning. Maybe it's just huge? Two minutes though...

The revert issue may have been fixed with the commit I just pushed. I noticed that it was still causing duplicates sometimes and figured out it was specifically when the range of the update was smaller than the range of the cached version. It's possible your revert deleted one or more rows which would trigger that bug.

[jvaldiviezo9]

Hi @SSlinky,

Thanks for the follow up. Regarding the repo, it's a personal repo, so I'd not be able to share it, but, It has some nesting folders. As I'm aware, current import/export solutions put the VBA files at folder level, and I'm doing that, all my VBA files are at the same directory level, so, there would not be a need to scan nested folders. Maybe the issue is that it's trying to scan nested folders, and in nested folders I have files that are related to Python packages, very large number of individual files. So, maybe the extension is trying to scan them?
Also, I have like 50+ files, some of them could surpass 1k lines of code, not sure if that could also make the scanning slower.

Thinking ahead, also, if the scanning is a possible issue, would it be nice to have a way to specify a PATH for the VBA code to scan to fetch definitions, and add the ability to include/exclude nested folders. In any case, I encourage you to check the "Naive Definitions" extension approach, on my tests it's very good and fast to pick Definitions, and also it can find References.

And, regarding the revert issue I guess will be available in the next update 1.7.5? I still have 1.7.4 as the latest version.

Thanks!

[SSlinky]

Yes, sorry, I haven't published a release for it because I have been working on other things at the same time. Hopefully soon it'll just appear to fix itself for you, but I'll try to remember to drop a note here so you get a notification.

With regards to reading files, it walks the entire tree (it has no way to know how people strucure a project) but it only reads files matching regex pattern /\.(cls|bas|frm)$/i. For reference, this is in /server/src/utils/helpers.ts.

/**
 * Recursively walks a directory structure and returns a map of file path to string content.
 * @param dirOrUri A directory as a path or 'file://' uri.
 * @param pattern A predicate to filter results.
 * @param files Used for internal recursive calls.
 */
export function walk(dirOrUri: string, pattern?: RegExp, files: Map<string, string> = new Map()): Map<string, string> {
	const logger = Services.logger;
	logger.debug(`Walking ${dirOrUri}`);

	// Walk the contents of the directory.
	const dir = dirOrUri.toFilePath();
	for (const name of fs.readdirSync(dir)) {
		const p = path.join(dir, name);

		// Check if we have a directory. This can occasionally throw at an OS level.
		let pIsDirectory: boolean | undefined;
		try { pIsDirectory = fs.statSync(p).isDirectory(); }
		catch (e) { logger.warn(`The OS threw an exception checking whether ${p} is a directory.`, 0, e); }
		if (pIsDirectory) {
			// Recursive call for directories.
			walk(p, pattern, files);
		} else if (pattern?.test(name) ?? true) {
			// Track files that match the pattern.
			logger.debug(`Found ${p}`, 1);
			logger.debug(`href: ${p.toFileUri()}`);
			let fileContent = '';
			try { fileContent = fs.readFileSync(p, 'utf-8'); }
			catch (e) { logger.error(`The OS threw an exception reading ${p}.`, 0, e); }
			files.set(p, fileContent);
		}
	}
	return files;
}

I'll have to have a look to see it can be optimised.

[jvaldiviezo9]

Don't worry, no rush, will be using the older version in the meanwhile, and thanks! :)

Now, regarding the scanning, as a suggestion in the future might work to have a setting where we can specify if we need to do a nested scan or just use the current directory, or an specific directory. Possibly, we could do that in settings.

Thanks again!