So this is a good point.  In terms of extending YSlow, it seems to me that
it would be possible to quantify (with some caveats) not only to what
degree you fit each of the criteria, but the likely performance improvement
if you did fully implement it.

So - I present specs for the YSlow Estimator.  Add it as a column in the
YSlow tab, it would take a couple seconds to run probably so don't
autopopulate it make them hit a button.

1.  Fewer HTTP requests.  You know how many concurrent requests that
browser's making and the overhead of each call; you should be able to
estimate the change in load time if you were e.g. downloading all 8 CSS
files as one.

2.  Use a CDN.  Should be able to hit a 20k reference image on Akamai and
calculate the speed difference between its d/l and the images in the page
d/l and show likely benefit.  css and js also.

3.  Expires header - trickier, no immediate idea.

4.  Gzip components - for components that weren't gzipped, gzip 'em, note
size difference, and multiply by the d/l speed they saw for that component
(time/size).

I won't go through all the rest - some are obvious (minify JS, avoid
redirects) and some are harder (JS at the bottom).  But I would think that
given a little analysis you could provide extremely valuable estimates not
only on "how good" someone's doing with respect to the arbitrary rating per
the 13 points, but actually show what their page performance could be if
they implemented each.  Naturally such an estimate would only be good for
the same browser in the same network location, but even so...

Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "Rakesh Rajan"
              <rakeshxp@...
              om>                                                        To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        Re: [exceptional-performance] Re:
              11/28/2007 12:19          YSlow + Minify JS + Dojo
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com






Vladimir,

Yups, I know that my JS code is optimized even though YSlow indicates it
otherwise.

What I trying to figure out was on how YSlow figures this out and wanted to
know if we can improve YSlow to take care of such scenarios. So YSlow could
a rating from 1-10 based on "how" ( not sure how to determine this )
compressed the code is ( rather than just a boolean: compressed or not) and
use it to compute the final score.

Cheers,
Rakesh

On Nov 28, 2007 11:40 PM, jeromakay <vladimir@...> wrote:


       Rakesh,

       In my opinion, I believe you shouldn't be so unflexible when it comes
       to YSlow and your code.

       Yslow does just some checks, benchmarks for your site. If you know
       your JS is minimified, but you have two three spaces and a comment
       inside your code, and you REALLY want that to stay there, why bother
       about YSLOW?

       My suggestion: You know it's minimified, move on! Yslow, it's just
       for
       refence and it's not really absolutely necessary to apply all those
       tips to get As all the time.

       Cheers!

       Vladimir Ghetau
       http://www.Vladimirated.com

       --- In exceptional-performance@yahoogroups.com, "Rakesh Rajan"



       <rakeshxp@...> wrote:
       >
       > I also have a query regarding Rule "Minify JS": All our JS code
       including
       > Dojo are compressed with the only exception that the compressed
       dojo js
       > contains a plani-text copyright. So this causes YSlow to flag it as
       not
       > compressed even though the JS file is compressed from 72.9 K to
       24.3 K.
       > Could some explain on how YSlow decides to flag a file as not
       > compressed/minified ?
       >
       > Thanks,
       > Rakesh

Estimating performance savings is difficult. In fact, the major
performance service companies have tried to build response time
simulators (person-months of effort) and gave up. Here are some of the
variables that make this a challenge:
   * What percentage of the users for web site X come in with an empty
cache vs. primed cache? This can vary 20% or more from one site to
another.
   * Where are most users located geographically?
   * What's the median bandwidth speed for users? (Can vary dramatically.)

This would be very cool if it could be done.

-Steve


--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
> So this is a good point.  In terms of extending YSlow, it seems to
me that
> it would be possible to quantify (with some caveats) not only to what
> degree you fit each of the criteria, but the likely performance
improvement
> if you did fully implement it.
>
> So - I present specs for the YSlow Estimator.  Add it as a column in the
> YSlow tab, it would take a couple seconds to run probably so don't
> autopopulate it make them hit a button.
>
> 1.  Fewer HTTP requests.  You know how many concurrent requests that
> browser's making and the overhead of each call; you should be able to
> estimate the change in load time if you were e.g. downloading all 8 CSS
> files as one.
>
> 2.  Use a CDN.  Should be able to hit a 20k reference image on
Akamai and
> calculate the speed difference between its d/l and the images in the
page
> d/l and show likely benefit.  css and js also.
>
> 3.  Expires header - trickier, no immediate idea.
>
> 4.  Gzip components - for components that weren't gzipped, gzip 'em,
note
> size difference, and multiply by the d/l speed they saw for that
component
> (time/size).
>
> I won't go through all the rest - some are obvious (minify JS, avoid
> redirects) and some are harder (JS at the bottom).  But I would
think that
> given a little analysis you could provide extremely valuable
estimates not
> only on "how good" someone's doing with respect to the arbitrary
rating per
> the 13 points, but actually show what their page performance could be if
> they implemented each.  Naturally such an estimate would only be
good for
> the same browser in the same network location, but even so...
>
> Ernest
> ______________________
> UN-altered REPRODUCTION and DISSEMINATION of
> this IMPORTANT information is ENCOURAGED.
>
>
>
>

>              "Rakesh Rajan"

>              <rakeshxp@...
>              om>
     To
>              Sent by:
exceptional-performance@yahoogroups
>              exceptional-perfo         .com

>              rmance@yahoogroup
     cc
>              s.com

>
Subject
>                                        Re: [exceptional-performance]
Re:
>              11/28/2007 12:19          YSlow + Minify JS + Dojo

>              PM

>

>

>              Please respond to

>              exceptional-perfo

>              rmance@yahoogroup

>                    s.com

>

>

>
>
>
>
> Vladimir,
>
> Yups, I know that my JS code is optimized even though YSlow indicates it
> otherwise.
>
> What I trying to figure out was on how YSlow figures this out and
wanted to
> know if we can improve YSlow to take care of such scenarios. So
YSlow could
> a rating from 1-10 based on "how" ( not sure how to determine this )
> compressed the code is ( rather than just a boolean: compressed or
not) and
> use it to compute the final score.
>
> Cheers,
> Rakesh
>
> On Nov 28, 2007 11:40 PM, jeromakay <vladimir@...> wrote:
>
>
>       Rakesh,
>
>       In my opinion, I believe you shouldn't be so unflexible when
it comes
>       to YSlow and your code.
>
>       Yslow does just some checks, benchmarks for your site. If you know
>       your JS is minimified, but you have two three spaces and a comment
>       inside your code, and you REALLY want that to stay there, why
bother
>       about YSLOW?
>
>       My suggestion: You know it's minimified, move on! Yslow, it's just
>       for
>       refence and it's not really absolutely necessary to apply all
those
>       tips to get As all the time.
>
>       Cheers!
>
>       Vladimir Ghetau
>       http://www.Vladimirated.com
>
>       --- In exceptional-performance@yahoogroups.com, "Rakesh Rajan"
>
>
>
>       <rakeshxp@> wrote:
>       >
>       > I also have a query regarding Rule "Minify JS": All our JS code
>       including
>       > Dojo are compressed with the only exception that the compressed
>       dojo js
>       > contains a plani-text copyright. So this causes YSlow to
flag it as
>       not
>       > compressed even though the JS file is compressed from 72.9 K to
>       24.3 K.
>       > Could some explain on how YSlow decides to flag a file as not
>       > compressed/minified ?
>       >
>       > Thanks,
>       > Rakesh
>

Hi, Ernest.

This server is not returning the correct Date value. As mentioned in
the article, I made the YSlow patch an option because "web servers
with incorrect Date values may cause inaccurate results." If you're
dealing with servers that don't return an accurate timestamp, feel
free to turn off this option. However, keep in mind that any site you
visit that contains stylesheets or scripts will suffer from the Net
Panel bug if you turn this option off.

The Net Panel bug is also explained in detail in the article
(http://developer.yahoo.com/yslow/netpanelpatch.html).

-Steve


--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
> But in this case, the YSlow version never shows my csses come down (have
> cleared cache, etc.).  They're in the CSS panel but just not the Net
panel.
>
> Ernest
> ______________________
> UN-altered REPRODUCTION and DISSEMINATION of
> this IMPORTANT information is ENCOURAGED.
>
>
>
>

>              "Steve Souders"

>              <souders@yahoo-in

>              c.com>
     To
>              Sent by:
exceptional-performance@yahoogroups
>              exceptional-perfo         .com

>              rmance@yahoogroup
     cc
>              s.com

>
Subject
>                                        [exceptional-performance] Re:

>              11/28/2007 11:47          Question on net panel

>              AM

>

>

>              Please respond to

>              exceptional-perfo

>              rmance@yahoogroup

>                    s.com

>

>

>
>
>
>
>
> Net Panel is not accurate in some cases. YSlow fixes the biggest
> problem, so what you're seeing with the YSlow patch enabled is _more_
> accurate. More info here:
>
> http://developer.yahoo.com/yslow/netpanelpatch.html
>
> -Steve
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> >
> > So I was investigating a slight bump we had in site performance,
and was
> > led astray briefly until I figured out what was going on - so with
> Firebug
> > + YSlow, I was hitting http://www.ni.com/company/management.htm and
> going
> > into the Net tab and seeing weird gaps; I figured out that the css'es
> > weren't showing up. I found the "Options... Patch in Net Panel from
> YSlow"
> > and unchecked it and then everything came in fine. I'm a little
> confused
> > about why the CSSes aren't showing in the YSlow version of the Net
panel
> > and I guess I didn't know that YSlow was replacing my Net panel in the
> > first place... What up? How do they differ?
> >
> > Thanks,
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
>

My hesitation is estimating the improvement in _response time_. I'm
not too optimistic about that. But, I _do_ think it would be great to
have a before vs. after estimator.

The "Stats" view in YSlow shows the number of HTTP requests and total
size for each category ("before" ). Below that we could show how that
summary would change if the performance suggestions were followed
("after"). For example, in the empty cache scenario you would go from
438K and 72 HTTP requests to 237K and 39 HTTP requests.

That's a very cool idea. I'm adding that to the list.

-Steve


--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
> I would say a tool would still be useful if it made base case
assumptions.
>
> 1.  No caching, or user specified percentage.
>
> 2.  Geographically - where you are
>
> 3.  Bandwidth - what you have
>
> In other words, start without worrying about theoretical cases - for
you, in your browser that you are running Firebug/Yslow, what would
the improvements be?
>
> Ernest
>
>
> ----- Original Message -----
> From: "Steve Souders" [souders@...]
> Sent: 11/28/2007 09:04 PM GMT
> To: exceptional-performance@yahoogroups.com
> Subject: [exceptional-performance] Re: YSlow + Minify JS + Dojo
>
>
>
>
> Estimating performance savings is difficult. In fact, the major
> performance service companies have tried to build response time
> simulators (person-months of effort) and gave up. Here are some of the
> variables that make this a challenge:
>   * What percentage of the users for web site X come in with an empty
> cache vs. primed cache? This can vary 20% or more from one site to
> another.
>   * Where are most users located geographically?
>   * What's the median bandwidth speed for users? (Can vary
dramatically.)
>
> This would be very cool if it could be done.
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> > So this is a good point.  In terms of extending YSlow, it seems to
> me that
> > it would be possible to quantify (with some caveats) not only to what
> > degree you fit each of the criteria, but the likely performance
> improvement
> > if you did fully implement it.
> >
> > So - I present specs for the YSlow Estimator.  Add it as a column
in the
> > YSlow tab, it would take a couple seconds to run probably so don't
> > autopopulate it make them hit a button.
> >
> > 1.  Fewer HTTP requests.  You know how many concurrent requests that
> > browser's making and the overhead of each call; you should be able to
> > estimate the change in load time if you were e.g. downloading all
8 CSS
> > files as one.
> >
> > 2.  Use a CDN.  Should be able to hit a 20k reference image on
> Akamai and
> > calculate the speed difference between its d/l and the images in the
> page
> > d/l and show likely benefit.  css and js also.
> >
> > 3.  Expires header - trickier, no immediate idea.
> >
> > 4.  Gzip components - for components that weren't gzipped, gzip 'em,
> note
> > size difference, and multiply by the d/l speed they saw for that
> component
> > (time/size).
> >
> > I won't go through all the rest - some are obvious (minify JS, avoid
> > redirects) and some are harder (JS at the bottom).  But I would
> think that
> > given a little analysis you could provide extremely valuable
> estimates not
> > only on "how good" someone's doing with respect to the arbitrary
> rating per
> > the 13 points, but actually show what their page performance could
be if
> > they implemented each.  Naturally such an estimate would only be
> good for
> > the same browser in the same network location, but even so...
> >
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
> >
> >
> >
>
> >              "Rakesh Rajan"
>
> >              <rakeshxp@
> >              om>
>     To
> >              Sent by:
> exceptional-performance@yahoogroups
> >              exceptional-perfo         .com
>
> >              rmance@yahoogroup
>     cc
> >              s.com
>
> >
> Subject
> >                                        Re: [exceptional-performance]
> Re:
> >              11/28/2007 12:19          YSlow + Minify JS + Dojo
>
> >              PM
>
> >
>
> >
>
> >              Please respond to
>
> >              exceptional-perfo
>
> >              rmance@yahoogroup
>
> >                    s.com
>
> >
>
> >
>
> >
> >
> >
> >
> > Vladimir,
> >
> > Yups, I know that my JS code is optimized even though YSlow
indicates it
> > otherwise.
> >
> > What I trying to figure out was on how YSlow figures this out and
> wanted to
> > know if we can improve YSlow to take care of such scenarios. So
> YSlow could
> > a rating from 1-10 based on "how" ( not sure how to determine this )
> > compressed the code is ( rather than just a boolean: compressed or
> not) and
> > use it to compute the final score.
> >
> > Cheers,
> > Rakesh
> >
> > On Nov 28, 2007 11:40 PM, jeromakay <vladimir@> wrote:
> >
> >
> >       Rakesh,
> >
> >       In my opinion, I believe you shouldn't be so unflexible when
> it comes
> >       to YSlow and your code.
> >
> >       Yslow does just some checks, benchmarks for your site. If
you know
> >       your JS is minimified, but you have two three spaces and a
comment
> >       inside your code, and you REALLY want that to stay there, why
> bother
> >       about YSLOW?
> >
> >       My suggestion: You know it's minimified, move on! Yslow,
it's just
> >       for
> >       refence and it's not really absolutely necessary to apply all
> those
> >       tips to get As all the time.
> >
> >       Cheers!
> >
> >       Vladimir Ghetau
> >       http://www.Vladimirated.com
> >
> >       --- In exceptional-performance@yahoogroups.com, "Rakesh Rajan"
> >
> >
> >
> >       <rakeshxp@> wrote:
> >       >
> >       > I also have a query regarding Rule "Minify JS": All our JS
code
> >       including
> >       > Dojo are compressed with the only exception that the
compressed
> >       dojo js
> >       > contains a plani-text copyright. So this causes YSlow to
> flag it as
> >       not
> >       > compressed even though the JS file is compressed from 72.9
K to
> >       24.3 K.
> >       > Could some explain on how YSlow decides to flag a file as not
> >       > compressed/minified ?
> >       >
> >       > Thanks,
> >       > Rakesh
> >
>

Right; I think that as long as the estimator is clearly just for your
circumstances it's an easy calculation to get improvement.  If you just
d/led that 100kb .js file at 5.0 kb/s (took 20s), well, then if it gets
minified to 50 kb you can honestly and accurately (enough for government
work) say that if you minified you would download it in 10s instead.  It's
an estimate only good for you in your current network location in your
browser with your settings - but it's still grotesquely useful.  Stick a
disclaimer on it saying "Not generalizable!!!  This is free!!!"

Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "Steve Souders"
              <souders@yahoo-in
              c.com>                                                     To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance]
              11/29/2007 01:33          improvement estimator [was Re:
              PM                        YSlow + Minify JS + Dojo]


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com







My hesitation is estimating the improvement in _response time_. I'm
not too optimistic about that. But, I _do_ think it would be great to
have a before vs. after estimator.

The "Stats" view in YSlow shows the number of HTTP requests and total
size for each category ("before" ). Below that we could show how that
summary would change if the performance suggestions were followed
("after"). For example, in the empty cache scenario you would go from
438K and 72 HTTP requests to 237K and 39 HTTP requests.

That's a very cool idea. I'm adding that to the list.

-Steve

--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
> I would say a tool would still be useful if it made base case
assumptions.
>
> 1. No caching, or user specified percentage.
>
> 2. Geographically - where you are
>
> 3. Bandwidth - what you have
>
> In other words, start without worrying about theoretical cases - for
you, in your browser that you are running Firebug/Yslow, what would
the improvements be?
>
> Ernest
>
>
> ----- Original Message -----
> From: "Steve Souders" [souders@...]
> Sent: 11/28/2007 09:04 PM GMT
> To: exceptional-performance@yahoogroups.com
> Subject: [exceptional-performance] Re: YSlow + Minify JS + Dojo
>
>
>
>
> Estimating performance savings is difficult. In fact, the major
> performance service companies have tried to build response time
> simulators (person-months of effort) and gave up. Here are some of the
> variables that make this a challenge:
> * What percentage of the users for web site X come in with an empty
> cache vs. primed cache? This can vary 20% or more from one site to
> another.
> * Where are most users located geographically?
> * What's the median bandwidth speed for users? (Can vary
dramatically.)
>
> This would be very cool if it could be done.
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> > So this is a good point. In terms of extending YSlow, it seems to
> me that
> > it would be possible to quantify (with some caveats) not only to what
> > degree you fit each of the criteria, but the likely performance
> improvement
> > if you did fully implement it.
> >
> > So - I present specs for the YSlow Estimator. Add it as a column
in the
> > YSlow tab, it would take a couple seconds to run probably so don't
> > autopopulate it make them hit a button.
> >
> > 1. Fewer HTTP requests. You know how many concurrent requests that
> > browser's making and the overhead of each call; you should be able to
> > estimate the change in load time if you were e.g. downloading all
8 CSS
> > files as one.
> >
> > 2. Use a CDN. Should be able to hit a 20k reference image on
> Akamai and
> > calculate the speed difference between its d/l and the images in the
> page
> > d/l and show likely benefit. css and js also.
> >
> > 3. Expires header - trickier, no immediate idea.
> >
> > 4. Gzip components - for components that weren't gzipped, gzip 'em,
> note
> > size difference, and multiply by the d/l speed they saw for that
> component
> > (time/size).
> >
> > I won't go through all the rest - some are obvious (minify JS, avoid
> > redirects) and some are harder (JS at the bottom). But I would
> think that
> > given a little analysis you could provide extremely valuable
> estimates not
> > only on "how good" someone's doing with respect to the arbitrary
> rating per
> > the 13 points, but actually show what their page performance could
be if
> > they implemented each. Naturally such an estimate would only be
> good for
> > the same browser in the same network location, but even so...
> >
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
> >
> >
> >
>
> > "Rakesh Rajan"
>
> > <rakeshxp@
> > om>
> To
> > Sent by:
> exceptional-performance@yahoogroups
> > exceptional-perfo .com
>
> > rmance@yahoogroup
> cc
> > s.com
>
> >
> Subject
> > Re: [exceptional-performance]
> Re:
> > 11/28/2007 12:19 YSlow + Minify JS + Dojo
>
> > PM
>
> >
>
> >
>
> > Please respond to
>
> > exceptional-perfo
>
> > rmance@yahoogroup
>
> > s.com
>
> >
>
> >
>
> >
> >
> >
> >
> > Vladimir,
> >
> > Yups, I know that my JS code is optimized even though YSlow
indicates it
> > otherwise.
> >
> > What I trying to figure out was on how YSlow figures this out and
> wanted to
> > know if we can improve YSlow to take care of such scenarios. So
> YSlow could
> > a rating from 1-10 based on "how" ( not sure how to determine this )
> > compressed the code is ( rather than just a boolean: compressed or
> not) and
> > use it to compute the final score.
> >
> > Cheers,
> > Rakesh
> >
> > On Nov 28, 2007 11:40 PM, jeromakay <vladimir@> wrote:
> >
> >
> > Rakesh,
> >
> > In my opinion, I believe you shouldn't be so unflexible when
> it comes
> > to YSlow and your code.
> >
> > Yslow does just some checks, benchmarks for your site. If
you know
> > your JS is minimified, but you have two three spaces and a
comment
> > inside your code, and you REALLY want that to stay there, why
> bother
> > about YSLOW?
> >
> > My suggestion: You know it's minimified, move on! Yslow,
it's just
> > for
> > refence and it's not really absolutely necessary to apply all
> those
> > tips to get As all the time.
> >
> > Cheers!
> >
> > Vladimir Ghetau
> > http://www.Vladimirated.com
> >
> > --- In exceptional-performance@yahoogroups.com, "Rakesh Rajan"
> >
> >
> >
> > <rakeshxp@> wrote:
> > >
> > > I also have a query regarding Rule "Minify JS": All our JS
code
> > including
> > > Dojo are compressed with the only exception that the
compressed
> > dojo js
> > > contains a plani-text copyright. So this causes YSlow to
> flag it as
> > not
> > > compressed even though the JS file is compressed from 72.9
K to
> > 24.3 K.
> > > Could some explain on how YSlow decides to flag a file as not
> > > compressed/minified ?
> > >
> > > Thanks,
> > > Rakesh
> >
>

Hmm, incorrect date?  Date on the servers and the load balancers seem to be
OK and NTP synced; what do you mean exactly?

(And it's a little weird this only affects our CSSes...)

Thanks,
Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "Steve Souders"
              <souders@yahoo-in
              c.com>                                                     To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance] Re:
              11/28/2007 03:47          Question on net panel
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com







Hi, Ernest.

This server is not returning the correct Date value. As mentioned in
the article, I made the YSlow patch an option because "web servers
with incorrect Date values may cause inaccurate results." If you're
dealing with servers that don't return an accurate timestamp, feel
free to turn off this option. However, keep in mind that any site you
visit that contains stylesheets or scripts will suffer from the Net
Panel bug if you turn this option off.

The Net Panel bug is also explained in detail in the article
(http://developer.yahoo.com/yslow/netpanelpatch.html).

-Steve

--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
> But in this case, the YSlow version never shows my csses come down (have
> cleared cache, etc.). They're in the CSS panel but just not the Net
panel.
>
> Ernest
> ______________________
> UN-altered REPRODUCTION and DISSEMINATION of
> this IMPORTANT information is ENCOURAGED.
>
>
>
>

> "Steve Souders"

> <souders@yahoo-in

> c.com>
To
> Sent by:
exceptional-performance@yahoogroups
> exceptional-perfo .com

> rmance@yahoogroup
cc
> s.com

>
Subject
> [exceptional-performance] Re:

> 11/28/2007 11:47 Question on net panel

> AM

>

>

> Please respond to

> exceptional-perfo

> rmance@yahoogroup

> s.com

>

>

>
>
>
>
>
> Net Panel is not accurate in some cases. YSlow fixes the biggest
> problem, so what you're seeing with the YSlow patch enabled is _more_
> accurate. More info here:
>
> http://developer.yahoo.com/yslow/netpanelpatch.html
>
> -Steve
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> >
> > So I was investigating a slight bump we had in site performance,
and was
> > led astray briefly until I figured out what was going on - so with
> Firebug
> > + YSlow, I was hitting http://www.ni.com/company/management.htm and
> going
> > into the Net tab and seeing weird gaps; I figured out that the css'es
> > weren't showing up. I found the "Options... Patch in Net Panel from
> YSlow"
> > and unchecked it and then everything came in fine. I'm a little
> confused
> > about why the CSSes aren't showing in the YSlow version of the Net
panel
> > and I guess I didn't know that YSlow was replacing my Net panel in the
> > first place... What up? How do they differ?
> >
> > Thanks,
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
>

The HTTP spec states that the Date response header "represents the
date and time at which the message was originated." It's currently Dec
4 (GMT), but some of the HTTP responses contain a Date header of Dec 3
(GMT). This is not the date the response was originated. Perhaps
there's a caching layer here that is returning the Date the cache was
updated.

-Steve


--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
>
> Hmm, incorrect date?  Date on the servers and the load balancers
seem to be
> OK and NTP synced; what do you mean exactly?
>
> (And it's a little weird this only affects our CSSes...)
>
> Thanks,
> Ernest
> ______________________
> UN-altered REPRODUCTION and DISSEMINATION of
> this IMPORTANT information is ENCOURAGED.
>
>
>
>

>              "Steve Souders"

>              <souders@yahoo-in

>              c.com>
     To
>              Sent by:
exceptional-performance@yahoogroups
>              exceptional-perfo         .com

>              rmance@yahoogroup
     cc
>              s.com

>
Subject
>                                        [exceptional-performance] Re:

>              11/28/2007 03:47          Question on net panel

>              PM

>

>

>              Please respond to

>              exceptional-perfo

>              rmance@yahoogroup

>                    s.com

>

>

>
>
>
>
>
> Hi, Ernest.
>
> This server is not returning the correct Date value. As mentioned in
> the article, I made the YSlow patch an option because "web servers
> with incorrect Date values may cause inaccurate results." If you're
> dealing with servers that don't return an accurate timestamp, feel
> free to turn off this option. However, keep in mind that any site you
> visit that contains stylesheets or scripts will suffer from the Net
> Panel bug if you turn this option off.
>
> The Net Panel bug is also explained in detail in the article
> (http://developer.yahoo.com/yslow/netpanelpatch.html).
>
> -Steve
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> > But in this case, the YSlow version never shows my csses come down
(have
> > cleared cache, etc.). They're in the CSS panel but just not the Net
> panel.
> >
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
> >
> >
> >
>
> > "Steve Souders"
>
> > <souders@yahoo-in
>
> > c.com>
> To
> > Sent by:
> exceptional-performance@yahoogroups
> > exceptional-perfo .com
>
> > rmance@yahoogroup
> cc
> > s.com
>
> >
> Subject
> > [exceptional-performance] Re:
>
> > 11/28/2007 11:47 Question on net panel
>
> > AM
>
> >
>
> >
>
> > Please respond to
>
> > exceptional-perfo
>
> > rmance@yahoogroup
>
> > s.com
>
> >
>
> >
>
> >
> >
> >
> >
> >
> > Net Panel is not accurate in some cases. YSlow fixes the biggest
> > problem, so what you're seeing with the YSlow patch enabled is _more_
> > accurate. More info here:
> >
> > http://developer.yahoo.com/yslow/netpanelpatch.html
> >
> > -Steve
> >
> > --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> > <ernest.mueller@> wrote:
> > >
> > >
> > > So I was investigating a slight bump we had in site performance,
> and was
> > > led astray briefly until I figured out what was going on - so with
> > Firebug
> > > + YSlow, I was hitting http://www.ni.com/company/management.htm and
> > going
> > > into the Net tab and seeing weird gaps; I figured out that the
css'es
> > > weren't showing up. I found the "Options... Patch in Net Panel from
> > YSlow"
> > > and unchecked it and then everything came in fine. I'm a little
> > confused
> > > about why the CSSes aren't showing in the YSlow version of the Net
> panel
> > > and I guess I didn't know that YSlow was replacing my Net panel
in the
> > > first place... What up? How do they differ?
> > >
> > > Thanks,
> > > Ernest
> > > ______________________
> > > UN-altered REPRODUCTION and DISSEMINATION of
> > > this IMPORTANT information is ENCOURAGED.
> > >
> >
>

Steve,

I'm reading your book and it says that Cache-Control was introduced in
HTTP/1.1 to be an alternative to Expires.

In the tool I have some warning for expiries tag  "These components do
not have a far future Expires header:"

Here is the response for my first image (I understand that it is here
b/c the Expires date is in the past)

Response Headers
Cache-Control: max-age=300
Content-Length: 3353
Content-Type: application/x-javascript
Expires: "Tue, 04 Dec 2007 08:57:50 GMT"
Pragma: cache
Server: barista/3.3.5
Via: Proxy/2,7,196269
Date: Tue, 04 Dec 2007 16:52:50 GMT
Proxy-Connection: Keep-Alive
X-lori-time-2: 1196787250781

On my second image the tool says I have "(no expires)" When actually I
have a Cache-Control property.  Why doesn't YSlow recognize the
Cache-Control?

Response Headers
Server: Microsoft-IIS/5.0
Cache-Expires: Tue, 04 Dec 2007 14:42:09 GMT
Accept-Ranges: bytes
Last-Modified: Fri, 09 Nov 2007 23:36:33 GMT
Etag: "8096585c2923c81:9f8"
Content-Length: 5904
Content-Type: image/gif
Cache-Control: max-age=1191
Date: Tue, 04 Dec 2007 16:52:48 GMT
Proxy-Connection: Keep-Alive
X-lori-time-2: 1196787248453

Another question I have which I didn't find in the book is what is the
difference b/t "Cache-Expires" and "Expires"?

Thanks

The expiration time has to be at least 48 hours in the future. In this
case it expires in 20 minutes so the benefit is reduced.

-Steve


--- In exceptional-performance@yahoogroups.com, "cbs4275"
<cbs4275@...> wrote:
>
> Steve,
>
> I'm reading your book and it says that Cache-Control was introduced in
> HTTP/1.1 to be an alternative to Expires.
>
> In the tool I have some warning for expiries tag  "These components do
> not have a far future Expires header:"
>
> Here is the response for my first image (I understand that it is here
> b/c the Expires date is in the past)
>
> Response Headers
> Cache-Control: max-age=300
> Content-Length: 3353
> Content-Type: application/x-javascript
> Expires: "Tue, 04 Dec 2007 08:57:50 GMT"
> Pragma: cache
> Server: barista/3.3.5
> Via: Proxy/2,7,196269
> Date: Tue, 04 Dec 2007 16:52:50 GMT
> Proxy-Connection: Keep-Alive
> X-lori-time-2: 1196787250781
>
> On my second image the tool says I have "(no expires)" When actually I
> have a Cache-Control property.  Why doesn't YSlow recognize the
> Cache-Control?
>
> Response Headers
> Server: Microsoft-IIS/5.0
> Cache-Expires: Tue, 04 Dec 2007 14:42:09 GMT
> Accept-Ranges: bytes
> Last-Modified: Fri, 09 Nov 2007 23:36:33 GMT
> Etag: "8096585c2923c81:9f8"
> Content-Length: 5904
> Content-Type: image/gif
> Cache-Control: max-age=1191
> Date: Tue, 04 Dec 2007 16:52:48 GMT
> Proxy-Connection: Keep-Alive
> X-lori-time-2: 1196787248453
>
> Another question I have which I didn't find in the book is what is the
> difference b/t "Cache-Expires" and "Expires"?
>
> Thanks
>

Thanks for the quick response!

Can this be set in the Cache-Expires or the Expires?  Does the browser
read these in a specific order?

-Chris

--- In exceptional-performance@yahoogroups.com, "Steve Souders"
<souders@...> wrote:
>
>
> The expiration time has to be at least 48 hours in the future. In this
> case it expires in 20 minutes so the benefit is reduced.
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, "cbs4275"
> <cbs4275@> wrote:
> >
> > Steve,
> >
> > I'm reading your book and it says that Cache-Control was introduced in
> > HTTP/1.1 to be an alternative to Expires.
> >
> > In the tool I have some warning for expiries tag  "These components do
> > not have a far future Expires header:"
> >
> > Here is the response for my first image (I understand that it is here
> > b/c the Expires date is in the past)
> >
> > Response Headers
> > Cache-Control: max-age=300
> > Content-Length: 3353
> > Content-Type: application/x-javascript
> > Expires: "Tue, 04 Dec 2007 08:57:50 GMT"
> > Pragma: cache
> > Server: barista/3.3.5
> > Via: Proxy/2,7,196269
> > Date: Tue, 04 Dec 2007 16:52:50 GMT
> > Proxy-Connection: Keep-Alive
> > X-lori-time-2: 1196787250781
> >
> > On my second image the tool says I have "(no expires)" When actually I
> > have a Cache-Control property.  Why doesn't YSlow recognize the
> > Cache-Control?
> >
> > Response Headers
> > Server: Microsoft-IIS/5.0
> > Cache-Expires: Tue, 04 Dec 2007 14:42:09 GMT
> > Accept-Ranges: bytes
> > Last-Modified: Fri, 09 Nov 2007 23:36:33 GMT
> > Etag: "8096585c2923c81:9f8"
> > Content-Length: 5904
> > Content-Type: image/gif
> > Cache-Control: max-age=1191
> > Date: Tue, 04 Dec 2007 16:52:48 GMT
> > Proxy-Connection: Keep-Alive
> > X-lori-time-2: 1196787248453
> >
> > Another question I have which I didn't find in the book is what is the
> > difference b/t "Cache-Expires" and "Expires"?
> >
> > Thanks
> >
>

Thanks for the quick response!

Can this be set in the Cache-Expires or the Expires? Does the browser
read these in a specific order?

-Chris

--- In exceptional-performance@yahoogroups.com, "Steve Souders"
<souders@...> wrote:
>
>
> The expiration time has to be at least 48 hours in the future. In this
> case it expires in 20 minutes so the benefit is reduced.
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, "cbs4275"
> <cbs4275@> wrote:
> >
> > Steve,
> >
> > I'm reading your book and it says that Cache-Control was introduced in
> > HTTP/1.1 to be an alternative to Expires.
> >
> > In the tool I have some warning for expiries tag "These components do
> > not have a far future Expires header:"
> >
> > Here is the response for my first image (I understand that it is here
> > b/c the Expires date is in the past)
> >
> > Response Headers
> > Cache-Control: max-age=300
> > Content-Length: 3353
> > Content-Type: application/x-javascript
> > Expires: "Tue, 04 Dec 2007 08:57:50 GMT"
> > Pragma: cache
> > Server: barista/3.3.5
> > Via: Proxy/2,7,196269
> > Date: Tue, 04 Dec 2007 16:52:50 GMT
> > Proxy-Connection: Keep-Alive
> > X-lori-time-2: 1196787250781
> >
> > On my second image the tool says I have "(no expires)" When actually I
> > have a Cache-Control property. Why doesn't YSlow recognize the
> > Cache-Control?
> >
> > Response Headers
> > Server: Microsoft-IIS/5.0
> > Cache-Expires: Tue, 04 Dec 2007 14:42:09 GMT
> > Accept-Ranges: bytes
> > Last-Modified: Fri, 09 Nov 2007 23:36:33 GMT
> > Etag: "8096585c2923c81:9f8"
> > Content-Length: 5904
> > Content-Type: image/gif
> > Cache-Control: max-age=1191
> > Date: Tue, 04 Dec 2007 16:52:48 GMT
> > Proxy-Connection: Keep-Alive
> > X-lori-time-2: 1196787248453
> >
> > Another question I have which I didn't find in the book is what is the
> > difference b/t "Cache-Expires" and "Expires"?
> >
> > Thanks
> >
>

I got a YSlow report for my site. In the Stats tab it shows,
0.0K 15 undefineds

What does 'undefineds' refer to? Can someone explain?

Thanks
Titus

With the new release of YSlow 0.9.0, it's possible for YSlow to find
HTTP requests that are not attached to any DOM elements. In this
situation, YSlow determines the "type" (image, js, css, etc.) by the
content-type response header and other clues. If a response can't be
classified (eg, it does not have a content-type response header) it is
labelled "undefined".

Can you tell us - was there a content-type response header? Do you
know what the response was in reality? Is there a URL we can look at?

-Steve


--- In exceptional-performance@yahoogroups.com, "sureshtytus"
<sureshtytus@...> wrote:
>
>
> I got a YSlow report for my site. In the Stats tab it shows,
> 0.0K 15 undefineds
>
> What does 'undefineds' refer to? Can someone explain?
>
> Thanks
> Titus
>

Hi Performancers!

Hope you're doing well with your site's tunning, I'm testing some code
I've build based on Steve's tips, and I want to try a real CDN besides
tweaking firefox to think that I'm actually on a real CDN :)

Anyone tried a service that's reliable, and .... afordable?

Thanks,

Vladimir Ghetau
http://www.Vladimirated.com/

I've recently heard suggestions for Panther Express
(http://www.pantherexpress.com/).

-Steve


--- In exceptional-performance@yahoogroups.com, "jeromakay"
<vladimir@...> wrote:
>
> Hi Performancers!
>
> Hope you're doing well with your site's tunning, I'm testing some code
> I've build based on Steve's tips, and I want to try a real CDN besides
> tweaking firefox to think that I'm actually on a real CDN :)
>
> Anyone tried a service that's reliable, and .... afordable?
>
> Thanks,
>
> Vladimir Ghetau
> http://www.Vladimirated.com/
>

Hi Steve,


Thanks for the suggestion, I'll check out their site.

After playing with it, I'll study how geographical locationed CND
content serving works. I'm on the right track thanks to your great info.

Ouch, Cal's from Flickr goes really well with it :)

Vladimir Ghetau
http://www.Vladimirated.com/

--- In exceptional-performance@yahoogroups.com, "Steve Souders"
<souders@...> wrote:
>
>
> I've recently heard suggestions for Panther Express
> (http://www.pantherexpress.com/).
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, "jeromakay"
> <vladimir@> wrote:
> >
> > Hi Performancers!
> >
> > Hope you're doing well with your site's tunning, I'm testing some code
> > I've build based on Steve's tips, and I want to try a real CDN besides
> > tweaking firefox to think that I'm actually on a real CDN :)
> >
> > Anyone tried a service that's reliable, and .... afordable?
> >
> > Thanks,
> >
> > Vladimir Ghetau
> > http://www.Vladimirated.com/
> >
>

Thanks Steve,

The URL I was looking at was the home page of
http://www.informationweek.com

Titus

--- In exceptional-performance@yahoogroups.com, "Steve Souders"
<souders@...> wrote:
>
>
> With the new release of YSlow 0.9.0, it's possible for YSlow to find
> HTTP requests that are not attached to any DOM elements. In this
> situation, YSlow determines the "type" (image, js, css, etc.) by the
> content-type response header and other clues. If a response can't be
> classified (eg, it does not have a content-type response header) it is
> labelled "undefined".
>
> Can you tell us - was there a content-type response header? Do you
> know what the response was in reality? Is there a URL we can look at?
>
> -Steve
>
>
> --- In exceptional-performance@yahoogroups.com, "sureshtytus"
> <sureshtytus@> wrote:
> >
> >
> > I got a YSlow report for my site. In the Stats tab it shows,
> > 0.0K 15 undefineds
> >
> > What does 'undefineds' refer to? Can someone explain?
> >
> > Thanks
> > Titus
> >
>

Hi, Titus.

Those are redirects. This is a bug in YSlow 0.9 that we'll fix now.
When doing Stats it's mistakenly calling redirects "undefined".

Thanks.

-Steve


--- In exceptional-performance@yahoogroups.com, "sureshtytus"
<sureshtytus@...> wrote:
>
> Thanks Steve,
>
> The URL I was looking at was the home page of
> http://www.informationweek.com
>
> Titus
>
> --- In exceptional-performance@yahoogroups.com, "Steve Souders"
> <souders@> wrote:
> >
> >
> > With the new release of YSlow 0.9.0, it's possible for YSlow to find
> > HTTP requests that are not attached to any DOM elements. In this
> > situation, YSlow determines the "type" (image, js, css, etc.) by the
> > content-type response header and other clues. If a response can't be
> > classified (eg, it does not have a content-type response header) it is
> > labelled "undefined".
> >
> > Can you tell us - was there a content-type response header? Do you
> > know what the response was in reality? Is there a URL we can look at?
> >
> > -Steve
> >
> >
> > --- In exceptional-performance@yahoogroups.com, "sureshtytus"
> > <sureshtytus@> wrote:
> > >
> > >
> > > I got a YSlow report for my site. In the Stats tab it shows,
> > > 0.0K 15 undefineds
> > >
> > > What does 'undefineds' refer to? Can someone explain?
> > >
> > > Thanks
> > > Titus
> > >
> >
>

That's a good topic; does anyone have any other CDN suggestions?  Akamai is
the big dog but there's competitors too, any leads?

Thanks,
Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "jeromakay"
              <vladimir@pixelto
              morrow.com>                                                To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance] cheap
              12/05/2007 09:41          CDN's - small business
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com






Hi Performancers!

Hope you're doing well with your site's tunning, I'm testing some code
I've build based on Steve's tips, and I want to try a real CDN besides
tweaking firefox to think that I'm actually on a real CDN :)

Anyone tried a service that's reliable, and .... afordable?

Thanks,

Vladimir Ghetau
http://www.Vladimirated.com/

Hmmm, well maybe I'm being dense, but it seems that regardless of Date
header, some things are showing up right in the and others not YSlow Net
panel  largely based on whether they're compressed.  We do use a Netscaler
for compression and caching, BTW.  But when I hit
http://www.ni.com/company/management.htm, I see:

Appear in YSlow Panel
management.htm    Fri, 07 Dec 2007 17:08:32 GMT
logo.gif                Fri, 07 Dec 2007 16:48:15 GMT
company_on.gif          Fri, 07 Dec 2007 16:27:05 GMT
white_right.gif         Fri, 07 Dec 2007 16:22:22 GMT

Don't Appear in YSlow Panel (but do in default Firebug net panel)
site.css          Fri, 07 Dec 2007 17:01:27 GMT
thecore.css       Fri, 07 Dec 2007 16:49:18 GMT
hierarchynavigation.css Fri, 07 Dec 2007 16:39:48 GMT
headerWidget.js   Fri, 07 Dec 2007 16:21:06 GMT

As you can see though the Date headers are "old" they are similarly old for
the images that do show up and the css/js that doesn't.  YSlow figures out
their download rates and all that just fine for the backdated images.

logo.gif headers
Age         1313
Date        Fri, 07 Dec 2007 16:48:15 GMT
Cache-Control     max-age=3600
Content-Length    2218
Connection  Keep-Alive
Via         NS-CACHE-7.0: 1
Etag        "KXMCBAABLLUKXMRR"
Server            Apache
Keep-Alive  timeout=300, max=967
Content-Type      image/gif

thecore.css headers
Age         1251
Date        Fri, 07 Dec 2007 16:49:18 GMT
Cache-Control     max-age=3600
Connection  Keep-Alive
Via         NS-CACHE-7.0: 1
Etag        "KXMCBAABLLXXWMRR"
Server            Apache
Keep-Alive  timeout=300, max=997
Content-Type      text/css
Content-Encoding  gzip
Content-Length    3315

The only real difference I can percieve in the headers between "shows up"
and "doesn't" is that our css and jses are compressed, and they're the only
things not showing up in the YSlow Net panel.  Are you sure this isn't a
problem with YSlow?

Thanks,
Ernest
______________________
UN-altered REPRODUCTION and DISSEMINATION of
this IMPORTANT information is ENCOURAGED.




              "Steve Souders"
              <souders@yahoo-in
              c.com>                                                     To
              Sent by:                  exceptional-performance@yahoogroups
              exceptional-perfo         .com
              rmance@yahoogroup                                          cc
              s.com
                                                                    Subject
                                        [exceptional-performance] Re:
              12/03/2007 06:32          Question on net panel
              PM


              Please respond to
              exceptional-perfo
              rmance@yahoogroup
                    s.com







The HTTP spec states that the Date response header "represents the
date and time at which the message was originated." It's currently Dec
4 (GMT), but some of the HTTP responses contain a Date header of Dec 3
(GMT). This is not the date the response was originated. Perhaps
there's a caching layer here that is returning the Date the cache was
updated.

-Steve

--- In exceptional-performance@yahoogroups.com, Ernest Mueller
<ernest.mueller@...> wrote:
>
>
> Hmm, incorrect date? Date on the servers and the load balancers
seem to be
> OK and NTP synced; what do you mean exactly?
>
> (And it's a little weird this only affects our CSSes...)
>
> Thanks,
> Ernest
> ______________________
> UN-altered REPRODUCTION and DISSEMINATION of
> this IMPORTANT information is ENCOURAGED.
>
>
>
>

> "Steve Souders"

> <souders@yahoo-in

> c.com>
To
> Sent by:
exceptional-performance@yahoogroups
> exceptional-perfo .com

> rmance@yahoogroup
cc
> s.com

>
Subject
> [exceptional-performance] Re:

> 11/28/2007 03:47 Question on net panel

> PM

>

>

> Please respond to

> exceptional-perfo

> rmance@yahoogroup

> s.com

>

>

>
>
>
>
>
> Hi, Ernest.
>
> This server is not returning the correct Date value. As mentioned in
> the article, I made the YSlow patch an option because "web servers
> with incorrect Date values may cause inaccurate results." If you're
> dealing with servers that don't return an accurate timestamp, feel
> free to turn off this option. However, keep in mind that any site you
> visit that contains stylesheets or scripts will suffer from the Net
> Panel bug if you turn this option off.
>
> The Net Panel bug is also explained in detail in the article
> (http://developer.yahoo.com/yslow/netpanelpatch.html).
>
> -Steve
>
> --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> <ernest.mueller@> wrote:
> >
> > But in this case, the YSlow version never shows my csses come down
(have
> > cleared cache, etc.). They're in the CSS panel but just not the Net
> panel.
> >
> > Ernest
> > ______________________
> > UN-altered REPRODUCTION and DISSEMINATION of
> > this IMPORTANT information is ENCOURAGED.
> >
> >
> >
> >
>
> > "Steve Souders"
>
> > <souders@yahoo-in
>
> > c.com>
> To
> > Sent by:
> exceptional-performance@yahoogroups
> > exceptional-perfo .com
>
> > rmance@yahoogroup
> cc
> > s.com
>
> >
> Subject
> > [exceptional-performance] Re:
>
> > 11/28/2007 11:47 Question on net panel
>
> > AM
>
> >
>
> >
>
> > Please respond to
>
> > exceptional-perfo
>
> > rmance@yahoogroup
>
> > s.com
>
> >
>
> >
>
> >
> >
> >
> >
> >
> > Net Panel is not accurate in some cases. YSlow fixes the biggest
> > problem, so what you're seeing with the YSlow patch enabled is _more_
> > accurate. More info here:
> >
> > http://developer.yahoo.com/yslow/netpanelpatch.html
> >
> > -Steve
> >
> > --- In exceptional-performance@yahoogroups.com, Ernest Mueller
> > <ernest.mueller@> wrote:
> > >
> > >
> > > So I was investigating a slight bump we had in site performance,
> and was
> > > led astray briefly until I figured out what was going on - so with
> > Firebug
> > > + YSlow, I was hitting http://www.ni.com/company/management.htm and
> > going
> > > into the Net tab and seeing weird gaps; I figured out that the
css'es
> > > weren't showing up. I found the "Options... Patch in Net Panel from
> > YSlow"
> > > and unchecked it and then everything came in fine. I'm a little
> > confused
> > > about why the CSSes aren't showing in the YSlow version of the Net
> panel
> > > and I guess I didn't know that YSlow was replacing my Net panel
in the
> > > first place... What up? How do they differ?
> > >
> > > Thanks,
> > > Ernest
> > > ______________________
> > > UN-altered REPRODUCTION and DISSEMINATION of
> > > this IMPORTANT information is ENCOURAGED.
> > >
> >
>

Hi folks!

Thanks for the excellent Yslow - its simply rocks.

With my latest yslow upgrade, I noticed that slowness hit the fan.

Yslow seems to try and load a movie that is streamed on my pages.

This makes FF throw the question (do you want to stop this script),
and Yslow seems to just wait for the movie to download fully on the
client, and the score by then, is terrible.

How can I avoid this?

If not, what's the downgrade process?

Any suggestions appreciated!

Cheers,

Hey all,

I've got a question regarding gzipping the components of your web
pages.  I'm new to this group and see most of the posts here relate to
the excellent YSlow!-extension, but since the group description
mentions 'web performance' in general, I though I'd share some
thoughts and questions on gzip.

We're using YSlow! trying to boost up our frontend performance and one
of the rules we're not getting an A at is 'Gzip Components'. We do
gzip our html pages, but don't gzip our css/js-files. I'm reading
everywhere a site caring about frontend performance should definitely
do this, and indeed, the file size improvements would be impressive. But;
I'm part of a development team for a large website (think 4+ billion
pageviews a month) and when we enabled gzip (through an Akamai CDN,
with Vary HTTP-headers), we started getting complaints from loads of
users, mentioning pages not showing up correctly because css/js-files
weren't getting through correctly. The issue seemed to relate to
Internet Explorer 6 SP1, that apparantely can't handle gzipped components.

Documentation of the bug we adressed this problem to, can be found here:
* http://support.microsoft.com/kb/823386/en-us.
Other related links:
* http://support.microsoft.com/?kbid=837251
*
http://dangermoose.blogspot.com/2006/01/gzip-compression-and-ie-do-not-mix.html

Now, if we're right - and this bug is indeed causing troubles for a
significant amount of our users - why isn't there any footnote about
this in articles that discuss frontend performance and/or gzip? I fail
to find lots of documentation + good solutions. For instance; I've got
the 'High Performance Web Sites' book by Steve Souders here with me
and it doesn't really mention anything related. (Or does it?)

At the moment this is really holding us back from using gzip, which is
a shame, because so many other users would benefit from enabling gzip.

Has this been an issue for others, too? How are other sites coping
with this problem? Seems to me sites like Yahoo / Google just gzip all
/ most of their content? Are they ignoring the users with IE6 SP1, or
are they serving them other content? Are we overlooking something?

Thanks!
Jurriaan

We actully pipe js/css via PHP, and use this to decide if use gzip,
this includes a few exceptions we found on the net somewhere (will try
to dig out the refernce at some point)

if (!empty($_SERVER['HTTP_ACCEPT_ENCODING'])) {
     $gzip = strstr($_SERVER['HTTP_ACCEPT_ENCODING'], 'gzip');
     $deflate = strstr($_SERVER['HTTP_ACCEPT_ENCODING'], 'deflate');

     $encoding = $gzip ? 'gzip' : ($deflate ? 'deflate' : 'none');

     if (!strstr($_SERVER['HTTP_USER_AGENT'], 'Opera') &&
             preg_match('/^Mozilla\/4\.0 \(compatible; MSIE
([0-9]\.[0-9])/i', $_SERVER['HTTP_USER_AGENT'], $matches)) {
         $version = floatval($matches[1]);

         if ($version < 6)
             $encoding = 'none';

         if ($version == 6 && !strstr($_SERVER['HTTP_USER_AGENT'], 'EV1'))
             $encoding = 'none';
     }
} else {
     $encoding = 'none';
}

With this have had no reported problems, (but only about 4million page
views a month, and about 25% of people still use IE6 to access the
site ;)

This of course requires a Vary: User-Agent header too.

HTH,

Barry

On Dec 11, 2007 11:02 AM, oemebamo <jurriaan@...> wrote:
> Hey all,
>
> I've got a question regarding gzipping the components of your web
> pages.  I'm new to this group and see most of the posts here relate to
> the excellent YSlow!-extension, but since the group description
> mentions 'web performance' in general, I though I'd share some
> thoughts and questions on gzip.
>
> We're using YSlow! trying to boost up our frontend performance and one
> of the rules we're not getting an A at is 'Gzip Components'. We do
> gzip our html pages, but don't gzip our css/js-files. I'm reading
> everywhere a site caring about frontend performance should definitely
> do this, and indeed, the file size improvements would be impressive. But;
> I'm part of a development team for a large website (think 4+ billion
> pageviews a month) and when we enabled gzip (through an Akamai CDN,
> with Vary HTTP-headers), we started getting complaints from loads of
> users, mentioning pages not showing up correctly because css/js-files
> weren't getting through correctly. The issue seemed to relate to
> Internet Explorer 6 SP1, that apparantely can't handle gzipped components.
>
> Documentation of the bug we adressed this problem to, can be found here:
> * http://support.microsoft.com/kb/823386/en-us.
> Other related links:
> * http://support.microsoft.com/?kbid=837251
> *
>
http://dangermoose.blogspot.com/2006/01/gzip-compression-and-ie-do-not-mix.html
>
> Now, if we're right - and this bug is indeed causing troubles for a
> significant amount of our users - why isn't there any footnote about
> this in articles that discuss frontend performance and/or gzip? I fail
> to find lots of documentation + good solutions. For instance; I've got
> the 'High Performance Web Sites' book by Steve Souders here with me
> and it doesn't really mention anything related. (Or does it?)
>
> At the moment this is really holding us back from using gzip, which is
> a shame, because so many other users would benefit from enabling gzip.
>
> Has this been an issue for others, too? How are other sites coping
> with this problem? Seems to me sites like Yahoo / Google just gzip all
> / most of their content? Are they ignoring the users with IE6 SP1, or
> are they serving them other content? Are we overlooking something?
>
> Thanks!
> Jurriaan
>
>
>
>
> Yahoo! Groups Links
>
>
>
>



--
Barry

- www.nearby.org.uk - www.geograph.org.uk -

Hello all,

I have updated YSlow today to 0.9.2 (previous version was 0.8).
After this, the following error message started to appear in firebug on
almost every page I visit:

Exception... "Component is not available" nsresult: "0x80040111
(NS_ERROR_NOT_AVAILABLE)" location: "JS frame :: file:///C:/Program%
20Files/Mozilla%20Firefox/components/nsSessionStore.js ::
sss_saveState :: line 1749" data: no

Does anybody know what might go wrong?

Thanx!