Searching GitHub repo only finds 2 of 26 files containing the pattern

[Liturgist]

When I search the remote GitHub repo from vscode it finds only two (2) files containing the pattern.

When I search the cloned repo on my local machine, Select-String finds twenty-six (26) files.

Should these produce the same results?

PS C:\> code --version
1.76.2
ee2b180d582a7f601fa6ecfdad8d9fd269ab1884
x64
PS C:\> code --list-extensions --show-versions
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]

[joyceerhl]

@Liturgist do you have a minimal repro (a sample repository and a specific pattern where this repros)? This sounds like a bug but I'd need more details in order to identify the cause.

[Liturgist]

@Liturgist do you have a minimal repro (a sample repository and a specific pattern where this repros)? This sounds like a bug but I'd need more details in order to identify the cause.

@joyceerhl, this is in an Enterprise repo. I am not sure that you will be able to access the repo. I do not think that I can/should just copy the files to a personal repo. Do you typically have access to Enterprise account repos?

[joyceerhl]

Do you mean a GitHub Enterprise repo? This extension (GitHub Repositories, for browsing repos without cloning locally) doesn't support opening repos hosted with GitHub Enterprise, so I'm not sure anymore what the scenario is here.

[Liturgist]

Do you mean a GitHub Enterprise repo? This extension (GitHub Repositories, for browsing repos without cloning locally) doesn't support opening repos hosted with GitHub Enterprise, so I'm not sure anymore what the scenario is here.

@joyceerhl, yes, this is an Enterprise account repo hosted by github.com for a Fortune < 10 company. This is not an on-prem GitHub Enterprise server.

It sounds like the extension will not search this repo remotely. Is that correct?

Can this repo be searched remotely with anything in Visual Studio?

[joyceerhl]

This is not an on-prem GitHub Enterprise server.

OK, that makes more sense.

the extension will not search this repo remotely

This extension does implement text search. I have a few questions:

• Is your search query a glob pattern? It's possible that the globs are being handled differently in this extension's search provider. If you could provide the specific glob pattern and examples of the 2 successful matches and the 24 unsuccessful ones that would be great (or provide an equivalent minimal repro if the matches or search contain anything proprietary).
• Are there any errors in the developer tools console (F1 > Toggle Developer Tools)? It's possible that the requests we make to GitHub's APIs to provide search results are failing (e.g. because of network proxies, transient errors, or perhaps a difference in API behavior between Enterprise and non-enterprise repos.
• To confirm, is this a GitHub Enterprise Cloud repo?

[Liturgist]

@joyceerhl. ok, I take it all back. After downloading the "index", the search results produces 29 files containing the pattern. Now I have to find out which extra files it found.

The "index" was several MiB. How different is that from cloning the repo?

Is there an easy way to get a list of the found files onto the clipboard?

One of the files has an icon to the left of its name that is orange and looks like a WiFi symbol, but the dot is in the lower left corner. What does that mean?

[Liturgist]

@joyceerhl, ok, things are a bit better. The three (3) extra files were .dtproj files and not .dtsx files that I was looking for.

What can I put in the "files to include" edit control to specify a base/root directory in the repo AND a file name extension? I want to search all *.dtsx files under the ./SSIS directory.

The WiFi-looking icon is gone. I found a way to copy the results list. I can get down to just the filenames with a couple of regex replacements.

[joyceerhl]

What can I put in the "files to include" edit control to specify a base/root directory in the repo AND a file name extension? I want to search all *.dtsx files under the ./SSIS directory.

Does SSIS/**/*.dtsx work for you?

The "index" was several MiB. How different is that from cloning the repo?

What we download for the search index is the git archive of the repo for the current commit (a regular git clone would
by default download your entire git commit history). This isn't necessary for just displaying the filetree, but for search it's valuable for working around some limitations of the current GitHub search API. Note also that GitHub Repositories supports searching/editing repos even in environments where git isn't available, e.g. browsers on mobile devices (via github.dev and vscode.dev).

One of the files has an icon to the left of its name that is orange and looks like a WiFi symbol

I'm not sure what this is, but if you see it again feel free to share a screenshot.

[Liturgist]

@joyceerhl, yes, SSIS/**/*.dtsx works. 26 files found. Thanks.

Code must not be using the GitHub search. I find that it will not identify patterns which are past column 1024.
https://github.com/orgs/community/discussions/52114

The icon is:

[joyceerhl]

I think that icon is just the icon for XML files in the Seti font (which is the default in VS Code).

Code must not be using the GitHub search. I find that it will not identify patterns which are past column 1024.

To clarify, do you mean that GitHub search doesn't find patterns past column 1024, or search in VS Code doesn't find patterns past column 1024? VS Code indexes the downloaded archive for search rather than using GitHub's search APIs directly in order to provide accurate search results.

[Liturgist]

The icon changes to be like the four-line icon above and below it. These are all .xml files.

GitHub search does not identify a pattern past 1024 characters. VS Code does. It identified all 26 files, just like Select-String.

[joyceerhl]

Thanks for clarifying. For the icon related issue, would you mind filing a bug at https://github.com/microsoft/vscode/issues and including a screen recording of what you are seeing there? If there is a bug with file icons I don't believe it would be specific to Remote Repositories.

I'm going to close this issue out since it sounds like searching GitHub repositories with this extension is working as expected.