311409 - Visit count improperly tracks accesses (doesn't increment) if the page forwards/redirects

Status: RESOLVED INCOMPLETE

(Camino Graveyard :: Bookmarks, defect)

Description

[Matt]

User-Agent:       Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b4) Gecko/20050914 Camino/1.0a1
Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8b4) Gecko/20050914 Camino/1.0a1

When I right click a bookmark in my bookmark bar, and press "get info," the
visit count says 0, even though I've been there (including by clicking it) many
times.

Reproducible: Always

Steps to Reproduce:
1.Right click bookmark in bookmark bar
2.Go to "visits" tab
3.
Actual Results:  
Said 0 vists

Expected Results:  
Said how many times I've visited it.

Comment 1

[Mike Pinkerton (not reading bugmail)]

interesting. wonder if this is a regression or if it ever worked right.

Comment 2

[Simon Fraser [no longer active]]

I can't reproduce. I drag a url to the bookmarks bar for a site I'm at, reload a
couple of times, and the visit count goes up.

Reporter: can you provide more detailed steps, including the exact URL in the
bookmark?

Comment 3

[Chris Lawson (gone)]

Works for me as well. Reporter, can you try a more recent nightly and see if you
still have the problem? The counter increments fine for me, no matter what means
I use to open the bookmark.

cl

Comment 4

[Matt]

Hey,

Sorry for the slow response, was out of town for a day. I'm using 1.0a1, but
I'll download the latest nightly build and let you know what happens.
-Matt

Comment 5

[Matt]

OK, here's what's going on. My Bookmark is to a site I login to-
http://vector.projectpath.com. When you go to that site, if you're not logged
in, you get the login screen: http://vector.projectpath.com/login. If you're
already logged in, you get the clients screen:
http://vector.projectpath.com/clients. When I make the link on the Bookmark bar
http://vector.projectpath.com, and click it, the visit count doesn't go up- I'd
assume this is because it is instantly refreshing to either /login or /clients.
If i make my bookmark bar link http://vector.projectpath.com/login, then it
works perfectly (assuming I'm not logged in). 

I'd say this is still a bug, because even if a page forwards or refreshes
instantly to another page, it should count as a visit. Especially when I'm
getting to the site by clicking the Bookmark bar link.

Sorry again for the delay in responding,
-Matt

Comment 6

[Chris Lawson (gone)]

We don't have any way of knowing in advance that the server is going to do the
redirect that it's doing (which, btw, appears to be done by the server, rather
than some kind of meta refresh), so we'd have an awfully hard time realising
what just happened and incrementing the counter.

I say this one's INVALID, because we're incrementing properly. It isn't our
fault that the bookmarked URL isn't the same as the redirected URL.

cl

Comment 7

[Matt]

Shouldn't the fact that I'm clicking that link in the bar, or even typing that
address in the location field, clue it in to up the counter though? It shouldn't
matter what happens AFTER the browser requests the page in my bookmark.

-Matt

Comment 8

[Simon Fraser [no longer active]]

(In reply to comment #7)
> Shouldn't the fact that I'm clicking that link in the bar, or even typing that
> address in the location field, clue it in to up the counter though? It shouldn't
> matter what happens AFTER the browser requests the page in my bookmark.

Well, this raises the question of what consitutes a "visit". We currently
increment the visit count every time you (re)load a page with a matching url, so
if a page auto-refreshes, the visit count will increase. I think this is pretty
bogus, and agree that a visit count based on button clicks/menu choices, and
perhaps typed urls, would make more sense.

Comment 9

[Chris Lawson (gone)]

Agree with Simon on that one.

Methinks the summary needs to be tweaked, then, since the bug isn't really that
bookmarks don't keep track of visits, but rather, they fail to keep track of
*accesses* (or "uses" or something; I'm not sure what term we should use here).
The difference in terminology is subtle but important.

Confirming.

cl

Comment 10

[Matt]

Summary updated, I'm definitely agreeing with you two. What's the next step on
this, and who decides if the change will be made or not?

(Sorry, this is my first bug-report experience).

-Matt

Comment 11

[Chris Lawson (gone)]

Either Simon or I (or possibly someone else who knows how to write code,
including you if you want!) can take a look at the code here and come up with a
patch. (The patching process may take quite a while, or it may take a few hours,
depending on how high the priority is.)

So now you get to sit back and wait. Endlessly. ;)

cl

Comment 12

[Chris Lawson (gone)]

Taking.

If I haven't posted a patch by the end of next week, someone e-mail me and
remind me.

cl

Comment 13

[Chris Lawson (gone)]

Anyone else have some input on what, exactly, should count as incrementing the
visit tracker?

Simon's comment 8 lists:

*button clicks
*menu selections
*typed URLs

I would append to this:

*autocompleted URLs, assuming the autocompleted URL is a bookmark

Anyone else?

How does History track URL visits? It seems like Bookmarks ought to be doing the
same thing, and if History already tracks accesses, is there any way we can get
that data and use it for Bookmarks?

cl

Comment 14

[Samuel Sidler (old account; do not CC)]

Unless there is a decision on what should happen and a patch posted soon, this will not make 1.0 and should not block it.

Retargeting for 1.1.

Comment 15

[Chris Lawson (gone)]

Oh, manually hitting cmd-R to reload should probably increment, too.

I agree (saying this again just for clarity) that auto-refreshing pages shouldn't increment.

Any other cases that should increment the counter? If not, I think I'm about ready to have a look at this for real.

cl

Comment 16

[Chris Lawson (gone)]

*** Bug 327545 has been marked as a duplicate of this bug. ***

Comment 17

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

*** Bug 338566 has been marked as a duplicate of this bug. ***

Comment 18

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Since no one's objected to the criteria in months, care to work up a patch, Chris?

-> 1.2, but we'll obviously take a reviewed patch

Comment 19

[Chris Lawson (gone)]

I dunno if the original page changed somehow, or if all the bookmarks changes in the past couple years did something, or maybe Core changed something, but the original testcase no longer works for me (i.e., it no longer exhibits this bug). Yes, I still have it bookmarked, and visiting it now increments the visit count.

It'd be nice to know if there are still cases where we don't increment the visit count properly. A testcase would be great if someone has time.

I think it's less of a problem that we *do* increment it for meta refresh tags, although if there's a way to determine that, we should still fix that part of the bug.