You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[BUG] Winget Detection does not work, however it is still listed in fastfetch -h packages-format #2434
I have checked the FAQ but the issue is not covered
Description
Winget Detection does not work. this is likely due to winget being overall a very slow package manager.
winget's list command is super slow and outputs unreliable results, you cannot rely on that.
in my fetch, i found a solution, winget looks in these registry keys for its packages:
"HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall*",
"HKLM:\SOFTWARE\WOW6432Node\Microsoft\Windows\CurrentVersion\Uninstall*",
"HKCU:\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall*"
these registry keys seem to be overall more reliable than winget's list command.
note: this includes apps NOT installed through winget directly, however theyre technically installed through winget if they are there
note2: while they are not always gotten through winget directly, winget's list command also shows them, and also sometimes has weird formatting.
while reading through winget documentation stuff, i figured out a way to find a SQLite database which logs installed winget stuff, its in %LOCALAPPDATA\Packages\Microsoft.DesktopAppInstaller_8wekyb3d8bbwe\LocalState\Microsoft.Winget.Source_8wekyb3d8bbwe\installed.db
thats what winget uses when you search winget list --source winget, so that could be what fastfetch would use rather than querying winget directly
I've read through the discussion and understand the concerns about relying on winget list due to its performance and inconsistent output. I also saw the suggestion about using the installed.db SQLite database and the request for more details.
I'm going to investigate the current package detection implementation, inspect the installed.db schema used by Winget, and evaluate whether it can be integrated without introducing unnecessary complexity or dependencies. If the approach looks suitable, I'll share my findings and open a PR.
If someone is already actively working on this issue, please let me know so I don't duplicate the effort.
Activity
CarterLi commentedon Jul 3, 2026
Winget is disabled by default, and can be enabled in config file with
{ "general": { "processingTimeout": 999999 }, "modules": [ { "type": "packages", "disabled": [] } ] }They contain a lot of garbages. I don't want to change the current logic before I clearly understand how winget handles that.
afterthoughttt commentedon Jul 5, 2026
oh, alright
winget does have a lot of unpredictable data and it barely keeps track of what it installs itself vs what it generically found
afterthoughttt commentedon Jul 5, 2026
while reading through winget documentation stuff, i figured out a way to find a SQLite database which logs installed winget stuff, its in %LOCALAPPDATA\Packages\Microsoft.DesktopAppInstaller_8wekyb3d8bbwe\LocalState\Microsoft.Winget.Source_8wekyb3d8bbwe\installed.db
thats what winget uses when you search winget list --source winget, so that could be what fastfetch would use rather than querying winget directly
CarterLi commentedon Jul 5, 2026
SQL please?
afterthoughttt commentedon Jul 5, 2026
wdym? im not super experienced with SQL stuff, it should be documented since winget-cli is OSS (https://github.com/microsoft/winget-cli/tree/master)
mannanloomba commentedon Jul 28, 2026
Hi! I'd like to work on this issue.
I've read through the discussion and understand the concerns about relying on
winget listdue to its performance and inconsistent output. I also saw the suggestion about using theinstalled.dbSQLite database and the request for more details.I'm going to investigate the current package detection implementation, inspect the
installed.dbschema used by Winget, and evaluate whether it can be integrated without introducing unnecessary complexity or dependencies. If the approach looks suitable, I'll share my findings and open a PR.If someone is already actively working on this issue, please let me know so I don't duplicate the effort.
CarterLi commentedon Jul 28, 2026
I dont think anyone is working on it.
RowanSkie commentedon Oct 17, 2026
Is this still being worked upon at all?
cerealdevv commentedon Oct 18, 2026
i don't think so, there's been no commits mentioning it