Closed GoogleCodeExporter closed 9 years ago
I should probably add that the header.php also includes the line (just below
<head>)
<base href="http://www.sellmyretro.com/" />
Original comment by rwap.services
on 3 Oct 2011 at 11:03
Well, if you have
<base href="http://www.sellmyretro.com/" />
<script type="text/javascript"
src="/themes/enuukGreen/js/jquery-ui-1.8.9.custom.min.js"></script>
then that should refer to the absolute URL
http://www.sellmyretro.com/themes/enuukGreen/js/jquery-ui-1.8.9.custom.min.js,
so the leading '/' shouldn't be necessary.
However, it should not be trying to lookup the ".pagespeed." URL from disk.
Do you have mod_pagespeed installed on http://www.sellmyretro.com/? Do you have
and ModPagespeedLoadFromDisk statements?
Original comment by sligocki@google.com
on 3 Oct 2011 at 3:35
Yes mod_pagespeed is installed on http://www.sellmyretro.com
I do have ModPagespeedLoadFromDisk statements :
ModPagespeedLoadFromFile "http://www.sellmyretro.com/themes/enuukGreen/"
"/var/www/html/themes/enuukGreen/"
Original comment by rwap.services
on 3 Oct 2011 at 6:17
Sounds like this is the same issue as 341.
I'm investigating. Are you actually getting 404s served out to publicly to
users? If so, removing that LoadFromFile directive should fix this for now.
Original comment by sligocki@google.com
on 3 Oct 2011 at 6:25
Aha, so on the example page:
http://www.sellmyretro.com/offer/details/1801
there is:
<base href="http://www.sellmyretro.com/"/>
...
<script type="text/javascript"
src="themes/enuukGreen/js/countdown/jquery.countdown.js.pagespeed.jm.cqsq3mj_NJ.
js"></script>
Which should point to
http://www.sellmyretro.com/themes/enuukGreen/js/countdown/jquery.countdown.js.pa
gespeed.jm.cqsq3mj_NJ.js (because of the base tag), but perhaps some browser or
spider thinks that the URL is
http://www.sellmyretro.com/offer/details/themes/enuukGreen/js/countdown/jquery.c
ountdown.js.pagespeed.jm.cqsq3mj_NJ.js
This is mystery number one. Do you happen to know which User-Agent's are
requesting these incorrect JS files?
Original comment by sligocki@google.com
on 4 Oct 2011 at 3:31
No idea as which user agents are doing this - if you look at issue 341 and the
excerpt from the access logs:
46.23.70.163 - - [02/Oct/2011:06:56:01 +0100] "GET
/store/eBay_Webstore//All+categories/Calculators/themes/enuukGreen/js/jquery-ui-
1.8.9.custom.min.js HTTP/1.1" 404 380 "-" "Serf/0.7.2
mod_pagespeed/0.9.18.7-900"
200.92.31.48 - - [02/Oct/2011:06:56:01 +0100] "GET
/store/eBay_Webstore//All+categories/Calculators/themes/enuukGreen/js/jquery-ui-
1.8.9.custom.min.js.pagespeed.jm.cF6ebxJV6w.js HTTP/1.1" 404 272
"http://www.sellmyretro.com/store/eBay_Webstore//All+categories/Calculators/Casi
o?lang=es_ES" "Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0;
MS-RTC LM 8; InfoPath.2; .NET CLR 2.0.50727)"
46.23.70.163 - - [02/Oct/2011:06:56:01 +0100] "GET
/store/eBay_Webstore//All+categories/Calculators/themes/enuukGreen/js/countdown/
jquery.countdown.js HTTP/1.1" 404 380 "-" "Serf/0.7.2
mod_pagespeed/0.9.18.7-900"
200.92.31.48 - - [02/Oct/2011:06:56:01 +0100] "GET
/store/eBay_Webstore//All+categories/Calculators/themes/enuukGreen/js/countdown/
jquery.countdown.js.pagespeed.jm.cqsq3mj_NJ.js HTTP/1.1" 404 272
"http://www.sellmyretro.com/store/eBay_Webstore//All+categories/Calculators/Casi
o?lang=es_ES" "Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0;
MS-RTC LM 8; InfoPath.2; .NET CLR 2.0.50727)"
It looks as though Mozilla/4.0 is one culprit (although the site displays ok in
Opera and Firefox)
Original comment by rwap.services
on 4 Oct 2011 at 4:01
Those User-Agents are actually for "MSIE 8.0", for historical reasons, almost
all User-Agents start with "Mozilla/4.0"
It turns out that there was a bug in IE8 which fits with these symptoms:
Thanks to Maks for great link:
http://blogs.msdn.com/b/ieinternals/archive/2009/07/27/bugs-in-the-ie8-lookahead
-downloader.aspx
So the conclusion is: This should not actually affect how any one sees the
page, it may cause slightly slower performance for people who are still using
buggy >1 year old versions of IE8. If this happens often enough it might be
worth considering, do you know how often it this happens?
Original comment by sligocki@google.com
on 4 Oct 2011 at 4:54
It does not happen that often - during the second half of September (the only
month I have data for at the moment) - around 1100 times out of 54,223
pageviews so probably can be ignored.
Original comment by rwap.services
on 4 Oct 2011 at 8:17
Summary was: Relative Paths to Javascript
I am defining this bug as whether or not we should optimize pages for the old
IE8 preloader which semi-ignores <base> tags.
If it is only 2% of users (and that number should be decreasing since it's been
fixed a year ago in IE8), it is probably not valuable to optimize for.
There is a separate issue, which is that we shouldn't be trying to load these
bogus .pagespeed. files from disk. That will be tracked in bug 341. I have not
been able to reproduce it yet.
Original comment by sligocki@google.com
on 4 Oct 2011 at 8:40
Original issue reported on code.google.com by
rwap.services
on 3 Oct 2011 at 11:02