All threads show up as read.

[~graffiti]

Is the CSS on the site messed up? I'm not getting bolded text for unread threads plus "updated" isn't listed on threads that have been updated.

But some of the ones I've already visited are a little lighter blue than the others.

[Bob_Peters]

JSS1138 wrote:

I should have been clearer.  My response was for the case where we can have one or the other.

Still a nonsense issue.  There must not be a situation where we can have one or the other.  Both speed and read tracking are MUST HAVES.  There can be no option to choose.

Do you see my point?

Of course I see your point.  However, the question was posed in such a way as to imply an either/or situation.

[Claudio_González1]

Bob, I also see his point. And the question was posed as an either/or situation: either keep the marginal speed increase gained by disabling Read Tracking, or get Read Tracking back and lose that speed increase.

Incidentally, I have seen no further official comments on this, and I am still not seeing Read Tracking in operation...

[pwillener]

It has been a week now, and it's still not fixed!  Is this yet another permanent improvement?

The forums are completely unusable this way; I have no idea what topics have been updated since my last visit.

[Claudio_González1]

I think that many people are not interpreting correctly what we are supossed to be voting. This was the question we were made, after being informed that Jive had turned Read Tracking off because of "performance issues":

adobe-admin wrote:

Read Tracking (identifying threads with new content since you last visit) does have an impact on performance.

Vote time. Which is more important? Read Tracking or Speed? The initial hope with read tracking is that it would do something similar to the old system that kept a high-water-mark for each forum so you could just pick up where you left off. But it never lived up to that.

I interpret this as: "If we turn RT on, speed will go down to what it was. Do you prefer the new increased speed, or RT turned on and the previous slower speed?".

Faced with this question, my answer is: the speed gain with RT off is marginal; not having RT is a nuissance. Therefore, I vote for RT.

[Niall_O_Donovan]

Thanks John!

[Claudio_González1]

Read Tracking is back! Frankly, I had given up all hope by now.

My thanks for whoever was responsible for this. And no, I don't see any noticeable difference in the download time of pages, regardless of what may be happening with the average downtime.

[Noel Carboni]

I do.  It's looking to me like its a few seconds slower now to do many things.  But I quite think it's worth it.

-Noel

[Claudio_González1]

Noel Carboni wrote:

I do.  It's looking to me like its a few seconds slower now to do many things.  But I quite think it's worth it.

-Noel

I am on a cable connection (not the fastest), and pages are taking between 2 and 3 seconds to load, which I would say about the same I got with RT off. Or, more clearly, if there is a difference, it is too small to detect using the seconds hand of a watch to measure download times, which is what I do.

Incidentally, times improved very much with my latest update of Firefox, which unfortunately was very little before RT was disabled (at most a couple of days), which has complicated my assessment of changes in speed. Before the update, pages in these forums took around 5 seconds to download.

[John Hawkinson]

Guys:

I am on a cable connection (not the fastest),

The speed of your network connection (cable, wireless G, etc.) really isn't too relevant here. Retrieving a typical short forum page doesn't require a lot of data transfer, so the speed of your access connection doesn't matter. Slowness will be dominated by the speed of the server's response. A typical such page is about 12k, though if you add in CSS, images, javascript, etc., it comes to 370k -- but almost alll of that will come from your browser's cache.

A full page thread is more like 30k, but again, not signficant.

I suspect the biggest issue in perceived slowness is how long it takes for the browser to get enough of the page such that it can begin rendering -- which is rather different from the ultimate load time. This is a human factors / psychology thing.

(for me, Firebug's Net panel says that a typical short page around 5 seconds to load, which seems way too long. But I'm not sure this is Jive's fault. The one I'm looking at sat for 2 second waiting for

http://wwwimages.adobe.com/www.adobe.com/adobeforums/resources/scripts/zapatec/styles/layouts/layout-common.css to load (not even on the Jive server!), and that's just a 745 byte file. I'm haven't run enough trials to make my data menaingful though. )

[Noel Carboni]

End to end latency of the internet connection DOES matter, however.  Nearly 70 requests are made of Adobe's servers for a forum page access, last I looked.  Some of these overlap, but many don't and each requires some time to complete.

A person with a connection with even a modest latency will see a few seconds wait to completely load a page here, as the request must be completely received by Adobe's servers, fulfilled, then the response must be completely received by the user's computer.

A person with lower latency will wait less.

Just for fun, start a CMD window and PING forums.adobe.com.

C:\TEMP>ping forums.adobe.com

Pinging adobe.hosted.jivesoftware.com [209.46.39.53] with 32 bytes of data:
Reply from 209.46.39.53: bytes=32 time=47ms TTL=234
Reply from 209.46.39.53: bytes=32 time=48ms TTL=234
Reply from 209.46.39.53: bytes=32 time=47ms TTL=234
Reply from 209.46.39.53: bytes=32 time=48ms TTL=234

As you can see, I'm lucky that my connection latency is reasonably low, so all those requests get through and I get my responses fairly quickly.  But what this means is that if Adobe's servers add more latency in fulfilling the requests, I will see it as a bigger jump in response time than someone who's already waiting longer.  It's just a bigger percentage, and that's what's perceptible.

The forum software is partially at fault in this...  70ish separate requests to display one web page is kind of a lot.

And, there's SO much script running that the speed of the script processor in your browser is also a significant factor.  IE9 is proving VERY fast at this.

-Noel

[John Hawkinson]

No, I do not think so. If you're on the other side of the planet, your latency will at worst be 100ms, and that difference is not significant here. And it should not "add up" -- multiple requests are in flight at the same time. And most importantly, nearly all of those 70 requests are cached and pipelined, making them equivalent to more like 4 requests.

Again, the network time is insignificant with respect to other factors, specifically the response time of the server before it gets the data onto the network.

Perhaps we should open another thread to continue this discussion, though I don't know if there's really much point.

[Noel Carboni]

More like 9 or 10 requests when you really look at it in action.

And, depending on how you have caching set, the server still has to respond and say "you've already got the latest page" (code 304).  But that's not true of most browsers at default settings.

And you're right about some of the requests being simultaneous.  But they're not all that way.  Bring up IE9's developer tools and watch a page display.  The Network tab gives you a nice Gantt chart.  You'll see long delays before some things are even requested.

Latency is only partially a distance factor.  Routers add latency as well.  I know mine recently went WAY down on a recent change of ISPs, and (last I looked) my house is in the same place it ever was.

-Noel

[Jochem van Dieten]

Noel Carboni wrote on 11/04/14 19:34:

Depending on how you have caching set, the server still has to respond and say "you've already got the latest page" (code 304).

One of the best examples is the default avatar for people who have not

set their avatar. Due to the way these forums are designed your browser

will still do a HTTP request to download it. For that reason I have

configured my Adblock plugin in Firefox with the following rule:

|http://forums.adobe.com/people//avatar/.png?a=-1

This blocks the default avatar. Together with blocking stats.adobe.com

and all the foresee survey Javascript that makes a noticeable impact.

And you're right about some of the requests being simultaneous. But they're not all that way.

The recommendation in the HTTP RFC is 2 simultaneous connections.

Traditionally browsers respect that, but recently some have increased it

to 6. Sending multiple requests over one connection simultaneously

(pipelining) is supported in the standard, but I am not aware of any

consistently working implementations.

Bring up IE9's developer tools and watch a page display. The Network tab gives you a nice Gantt chart. You'll see long delays before some things are even requested.

Usually that is Javascript processing. Despite all the improvements,

Javascipt is still single-threaded and when Javascript runs everything

else stops. That is why the recommendation is to put Javascript at the

bottom of the page: the Javascript runs while the content is already

visible for the user.

Latency is only partially a distance factor. Routers add latency as well. I know mine recently went WAY down on a recent change of ISPs, and (last I looked) my house is in the same place it ever was.

That is not so much a router issue as a network protocol issue. Most

networks still have some ATM layer in the last mile and in many DSL

lines that ATM layer is still configured to be interlaced. Gives higher

nominal throughput especially when the line is nearing capacity, but

will cost some additional latency. Once such an ATM step can cost more

latency as 10 routers.

[Jim_Simon]

Now what!  It's off again!

This is very, very frustrating.

[Noel Carboni]

Seems to work OK here, and I didn't notice any outages today (though to be fair there were some stretches of hours I didn't look).

-Noel

[the_wine_snob]

Mine is also working fine in Chrome, though there was a forum outage at ~ 11:00AM PDT, Thursday, April 21, 2011.

It lasted until ~ 2:30PM, PDT, in one form, or another.During that outage, there were several log-outs, when I could get a server, and the forums ran at about 1/1000 speed. Gave up, and came back later.

Hunt

[Jim_Simon]

Hmm.  Seems as if something changed, though.  While the "Updated' message does appear when using IE9, my Stylish setting which turned them into Red Flags in Firefox isn't working any more.  I get no indication of the message being updated at all.

Something changed.

[Jim_Simon]

OK. Found the problem.  The script was using a GIF image hosted on  Adobe's site to replace the "updated" text.  Adobe changed their URLs  around, so the GIF is no longer where it was.

[Noel Carboni]

Can you please screen grab what you've got set up and post it here?  Sounds interesting.

I'm just seeing threads with new content as bold myself.

-Noel

[the_wine_snob]

Yeah, I miss my "red flags" from the old Forum (the one that was not broken... ).

Hunt

[Jim_Simon]

Here's what I see using Firefox and Stylish.  Honestly, I don't know how anyone could stand the default look of these forums.  I'd just leave if I had to put up with that.

[the_wine_snob]

Ah, the "red flags." That alone might be enough for me to change to FF. I miss the flags.

Hunt

[Guest]

nevermind!

[Noel Carboni]

JSS1138 wrote:

Honestly, I don't know how anyone could stand the default look of these forums.  I'd just leave if I had to put up with that.

Out of curiosity, what do you feel is so much better about the view you're seeing than this?  The bold subjects may be even more visible than your little red flags.  I admit, you'll need less scrolling...

-Noel

[the_wine_snob]

On my 17" high-rez laptop, the "(Updated)" is a bit less obvious, than the red flags.

One forum, that I frequent has another nice feature.

If one has not visited a thread, then all replies show fully expanded.

If one re-visits, only those unread replies show expanded, but all collapsed threads can be expanded, one by one, or all, in Expand All.

That allows one to also view in Threaded Mode, so that it's easy to see who is replying to whom.

Just some observations from other forums.

Hunt