URL that returns latest stable version of EdgeDriver does not seem to take OS, CPU arch and major browser version into account

[gburghardt]

I honestly do not know how to quantify this. It is either the opening for a discussion about the difficulty of auto-updating EdgeDriver, or the URL that returns the "latest stable" version of EdgeDriver just isn't useful anymore.

Issue 168 in the WebDriverManager.Net project on GitHub reports that auto-updating EdgeDriver is broken. Their current logic for getting the "latest" driver goes as follows:

1. Get the version number returned by https://msedgedriver.azureedge.net/LATEST_STABLE
2. Using a pattern in the URL, get the zip file (either win32 or win64).

The problem is that the version number (as of right now) returns 103.0.1264.51. There is no win32 or win64 version of this driver. That version number is only valid for MacOS and Linux. While there is nothing inherently wrong with this, it does present problems when attempting to update EdgeDriver automatically. It means the "LATEST_STABLE" URL returns the latest stable version without respect to the operating system.

Is this a bug in the URL endpoint to return the latest stable version of EdgeDriver, or is this the intended behavior?

Do we just need to replumb the logic for inferring the latest EdgeDriver by downloading the 2.5MB XML file at https://msedgedriver.azureedge.net/ and parsing through it, thereby ignoring the LATEST_STABLE URL?

I guess I'm just looking for a little direction on how we should be using these public URLs to infer which EdgeDriver supports a particular version of Edge.

A valid answer could also be "this is not the place to request this kind of help" (but it would be nice to know where to submit this feedback).

---

Full disclosure, the logic for auto-updating ChromeDriver is suffering from a similar kind of issue, which of course is not something Microsoft has any control over.

AB#42619861

[bwalderman]

Hi @gburghardt. I actually just responded to a similar issue only a few minutes ago: #39

The LATEST_STABLE endpoint is indeed not very useful in its current state. Hopefully the info in that thread is helpful.

[bwalderman]

Adding this to our internal backlog. This will likely be useful for tools like WebDriverManager and Selenium Manager.

[todayumay]

Hi guys,

I faced with the same issue (please link the #175 issue here) - is there any details on the issue roots?
Could you please clarify the purpose of /LATEST_STABLE Endpoint? What exact platform it's point to? I found out that it doesn't actually point to the latest build, as the MacOS driver build '130.0.2849.81' is newer than one it responds with ('130.0.2849.80').