Re: Problem with YouTube videos if cookies are disabled
Posted: Sun Dec 11, 2011 10:40 am
@Ian - Good. Now the bug is clear, I think it's up to the mods to maybe move this thread to the development forum for the dev's attention?
Not sure exactly what the tests are, but it's definitely not a single data-stream, as the speedometer fluctuates during the test period. I take that to mean that they're throwing different types of d/l and u/l material at the machine. It's a service branded by my ISP, but is based on http://www.ookla.com/speedtest (their logo shows on the page). If you want to attach your packet sniffer or whatever and examine the test material, I'm sure we'd all love to see exactly what types of things they're benchmarking.dhouwn wrote:You are talking about measuring linear line speed, i.e. how quick a single data-stream runs? If yes than that's no surprise, it would been quite an issue if Firefox somehow limited this.Tom T. wrote:Have seen claimed improvements in speed. Benchmarking all three with speedtest, at about the same time of day (to avoid different levels of ISP, Internet, and server congestion) shows no statistically-significant difference.
It is for medhouwn wrote:Not sure I agree, NoScript is (i.a.) about controlling JS but not necessarily about running "as little JS as possible".We're here because we allow as little JS as possible.
, but let me restrict the generalization somewhat: We're here because we don't want to allow every JS on the planet. Most sites come with so much 3rd-party garbage, ranging from useless to annoying to privacy-invasive to possibly dangerous, that even a somewhat more relaxed attitude toward JS is still going to reduce the total amount of JS the browser has to render, versus without NoScript or some other (usually, less selective) form of blocking. So I still maintain that the speed value of faster JS rendering, while real (because we need it for our trusted or TA'd sites) is greatly diluted by a selective approach as to what scripting to allow.
btw, I'm a firm believer in the Principle of Least Privilege. Even if there isn't a safety risk (how can you be sure?), why waste the bandwidth, memory, CPU cycles, etc.?
More speed "improvement" comes from link-prefetching, and also from speculative HTML parsing in Fx 4+, both of which may be significant privacy risks.
Some day, I'll get a stopwatch and try to compare page-load times, etc. on all three branches. (Or not.) IMHO, the other factors: congestion of ISP, Internet pipelines, and destination server are enough to hide any perceptible gains. Admittedly, my usage is probably lower than the average user's: I''m on the computer many hours a day, but doing relatively low-resource things, like e-mail, this forum... I'm not an online gamer, don't d/l feature-length movies, etc. Which is a good reason for Fx to maintain the two branches: whatever they can safely do for the gamers and movie fans, while keeping the simpler version in which YouTube works without all of the hassles described here.dhouwn wrote:There is also DNS-prefetching, quite a lot of networking improvements (see http://www.browserscope.org/?category=network) like changes allowing for more resources to be loaded simultaneously. Also things that might improve latency like forcing Nagle's algorithm off. Also, caching for SSL content.
IMHO. YMMV. Color rendering is said to be improved. OK. On a laptop, I can't really tell the difference.
I'm talking about everything I see on the screen. ... and while we're on the subject of color, gray balloons on a gray background, and black text on a dark gray background, are *really* bad choices, assuming that they wish to appeal to anyone over 30.dhouwn wrote:Huh? Are you talking the iCC profiles in images? Well, I would guess that you won't find that many images on the web that are displayed in Firefox with colour correction for various reasons (AFAIK, Firefox just has some half-heartedly support for a ICCv2).
F2 and F3 were much easier on the eyes for that demographic. Footprint up about 50% from 2 to 3, been more than doubled from 2 to 8. For what?
I used "footprint" in terms of "HD space consumed", and to a slightly lesser degree, number of files, each of which requires a 1k entry in the MFT, and adds to backup time and size. Or adding a 10+MB new file, xul.dll, in F3, while F2 seemed to get along just fine without it.dhouwn wrote:Mostly JS I would guess (currently my about:memory tells me that JS accounts for about half of all explicit allocations). Other reasons might include more aggressive caching (sometimes less though) and if you have it hardware acceleration if you have it enabled. Not sure about the HTML5 parser/dom but I can't exclude it also consuming more than the old ones (I remember reading a blog post from an Opera developers who talked about this how it did in Opera's case).
BTW, I believe you might see even lower memory usage with Firefox 1.5 and 3, former because it does less caching (which is why Firefox 2 was considered a memory hog at its time) and latter because they actively tried to the reduce the memory usage compared to Firefox 2.
Caching is user-configurable, so no complaint there.
[rant]I just see very little *tangible, visible* benefit from Fx 4+, with many negatives. Like this thread and YouTube... and many others. [/rant]
Problem solved.therube wrote:FF 3.6, FF 2.x, aren't doing HTML5 (are they), so using those to test shows nothing (& you would expect that Youtube would fall back to Flash in those cases)
I haven't actually seen any there, though my visits are not frequent. If you can point me to one in another format, I'll be happy to do a test case. But they don't do Apple QuickTime, do they? ... what other formats are there at YT?therube wrote: & the OP was talking about clips in other then Flash format.
.tlu wrote:Not here. With youtube.com and ytimg.com whitelisted and cookies blocked (with CookieMonster) all videos play without a problem
My point was that on Fx 3.6, those scripts can remain blocked, along with cookies, and there is *still no issue*. Fx 8.01 is what complicated things.
See above reply to dhouwn re: in general, the less allowed, the better, IMHO.
@ Ian: With Fx 3.6, there's no trade-off to make. We can have our cake and eat it, too. But for Fx Newer, it quickly becomes second-nature to open the menu, point to BO, and recognize the pattern, as mentioned. BTW, do you have NS Menu set to open on mouseover versus having to click to open? Two seconds to get the video playing, without the cookie.
However, I don't think the cookie method is less safe; only a privacy issue. That privacy issue can be removed by deleting the cookie before going elsewhere. Either manually; using various add-ons (tlu mentioned Cookie Monster); close/restart the browser; maybe write a little batch script that empties cookies with two clicks, (I did that for LSO's, a/k/a Flash Cookies, a/k/a "Personally Identifiable Elements", a/k/a "PIE", a nice pun on "Cookie", because there was some discontent with the add-on that did this.) or other ways.
Cheers - Tom
All answers appreciated. TIA.