935673 - Firefox landing page incorrectly detects the previous version as the latest version

Status: VERIFIED FIXED

(www.mozilla.org :: Pages & Content, defect)

Description

[jerryj]

Attachment: ff-noupdate.jpg

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/30.0.1599.101 Safari/537.36

Steps to reproduce:

run FF 24.0 -- about says update available, click on link


Actual results:

says no update is available


Expected results:

update

Comment 3

[Kohei Yoshino [:kohei]]

I think this can be fixed by just removing this line:
https://github.com/mozilla/bedrock/blob/master/media/js/firefox/new.js#L21

Comment 4

[Kohei Yoshino [:kohei]]

Attachment: Pull Request on GitHub

Comment 5

[Chris More [:cmore]]

This is on purpose for people who have not restarted their browser from a silent update yet and are running latest version minus 1. I vote that we keep this as-is until we have a way to detect if a silent update is applied and just waiting for an update. This has been discussed many times in the past few months and SUMO is doing latest version minus 1 too. This was also discussed in bug 883800.

Comment 6

[Drew Willcoxon :adw]

You can reproduce comment 0 without any silent updates, though.  Just download Firefox 24, start it, and open the about window.  Surely that's not right.

Comment 7

[Chris More [:cmore]]

(In reply to Drew Willcoxon :adw from comment #6)
> You can reproduce comment 0 without any silent updates, though.  Just
> download Firefox 24, start it, and open the about window.  Surely that's not
> right.

24 should prompt you to update since latest version - 1 = 25 and that's not < 24. I can try manually downloading Firefox 24 and testing it, but when this bug was submitted a month ago, the latest version was 25. 25-1=24 and that's the scenario that we don't prompt users to update.

As mentioned in the other bug, we are unable to detect if a silent update has been applied or if a silent update will soon be applied as silent updates are slowly throttled out to everyone around the world. Instead of having to have users manually update or update while a silent update is already applied, we only tell the user to manually update when they are more than one version old. We are thinking that we may change this logic to be only true for the first week after a new release goes out to give time for the silent updates to be throttled around the world and after users have more likely restarted once.

Comment 8

[Roman]

(In reply to Chris More [:cmore] from comment #7)
> Instead of
> having to have users manually update or update while a silent update is
> already applied, we only tell the user to manually update when they are more
> than one version old.

Then this should be applied consistently on the website and in firefox itself.

The experience I have is:

* firefox nudges me to update
* I go to firefox.com to download the new version (automatic updates are disabled because firefox is installed in a non-user-writable location)
* The website tells me that I'm using the latest version!

Besides just being confusing, it makes much harder for me to actually download the new version

Comment 9

[jerryj]

How embarrassing that something as simple as a KNOWN “off by one” error is still not fixed after all of this time (now going to 40 days).

Here I am running 25.0.1, go into the ‘about’ dialog, which tells me that updates are available, click on the link, and am told “Congrats! You’re using the latest version of Firefox -- Different by design” (that behavior truly is ‘different’, but not in a good way).

Maybe this needs to be escalated to a ‘security’ level, since it prevents any security fixes (in the latest release) from being installed.

Comment 10

[Chris More [:cmore]]

Ibai: Can you provide some feedback here from what you have heard on SUMO on the logic? I know that you changed the logic on SUMO to the same latestVersion-1.

Jerry/Roman: It is not an error as described in the comments as this is completely expected behavior even if it is not consistent with the product itself. It is not fully idea and we are working on other possible solutions. We were finding that users were downloading and installing the latest version over top of the latest version quite a bit (pave overs) without knowing that silent updates exist. That's a lot of wasted bandwidth and time on the end user's perspective to install the latest version over top of the latest version. If users want to pave over, we were sending them to the /firefox/all/ page that has links to all of the builds for all of them. We have a link on the /firefox/new/ page that says "Download a fresh copy" and that is not preventing you from downloading the latest Firefox version. 

In summary, removing the download button is not preventing users from downloading as there is a link right on the page that says "Download a fresh copy". I also understand the logic may be confusing and not apparent to users and we should improve that.

Here are a few options:

1) remove the latest version - 1 logic and just make it latest version.

2) Change the "Download a fresh copy" to the download itself instead of /firefox/all/.

3) Change /firefox/all/ to detect the user's language and provide the download for their language at the top of the list and then provide options to download any other language.

Comment 11

[Ibai Garcia [:ibai]]

We really don't have any positive/negative feedback on this.

What we know though is that no matter what, users will try to download Firefox, even if this is the same version are the one that they already have. We should try to help them make this decision correctly, although it seems that doing this we add more frustration than helpfulness if we don't communicate it properly.

Comment 12

[Kohei Yoshino [:kohei]]

I still stand with removing the -1 logic because it's rather confusing users. I think we don't have to care much about pending silent updates. There's no way to detect such a state anyway.

ESR makes the situation difficult. 24.1.x cannot be detected since the UA strings has been kept as 24.0 (See Bug 870540). We may have to check navigator.buildID instead. It's complicated, platform-dependent but reliable than UA, IMO.

Comment 13

[Chris More [:cmore]]

Habber: are you fine with removing the latest version - 1 logic on the download page and prompting users to update if they aren't on the latest version? The -1 logic seems to be confusing especially when the product isn't using that same logic. It is a really quick fix and doesn't affect anything else.

Comment 14

[Holly Habstritt Gaal [:Habber]]

Given the particular flow that Jerry mentioned, I agree that we should remove the logic here. 

However, I'd like to renew our request to add the ability to know if a silent update has occurred. We want to catch users who already have a new version of Firefox installed and stop them from downloading fresh copy.

Comment 15

[Chris More [:cmore]]

Kohei: Please proceed with a PR to remove the -1 and you can reference comment 14 as approval. 

Habber: Let's file a bug in https://bugzilla.mozilla.org/enter_bug.cgi?product=Toolkit&component=Application%20Update about seeing if mozilla.org will ever be able to tell users a silent update has been applied, but a restart is needed.

Comment 16

[Kohei Yoshino [:kohei]]

Comment on attachment 8345890 [details] [review]
Pull Request on GitHub

Reopened the pull request.

Comment 17

[[github robot]]

Commits pushed to master at https://github.com/mozilla/bedrock

https://github.com/mozilla/bedrock/commit/3a24576228ffae4bb94a6517091bb75add75460e
Bug 935673 - Firefox landing page incorrectly detects the previous version as the latest version

Just remove the subtraction. It has rather confused users.

https://github.com/mozilla/bedrock/commit/5c482c86e322eb6154bb20a24dafc359bbaaee3a
Merge pull request #1514 from kyoshino/bug-935673-wrong-latest-version

Bug 935673 - Firefox landing page incorrectly detects the previous version as the latest version

Comment 18

[Mike Alexis [:malexis]]

This has been pushed. Need to verify working as expected on prod.

Comment 19

[Mike Alexis [:malexis]]

Raymond can you verify?

Comment 20

[raymond [:retornam] (inactive)]

fixed on production http://www.mozilla.org/en-US/firefox/new/ tested using Firefox 26 and Firefox Aurora 23.0a2 (2013-06-22)