TSGL: IE refresh latency--explanation?

[H Davis]

Don,

I have 5 web sites which have been on 2 different hosting services (all 
on the same one now) and the new page shows up immediately on my browser 
and I don't have to do any cache clearing to see it. If the page I'm 
updating is already visible on my browser all I have to do is click the 
refresh button and I'll see the new page. The new page is there as soon 
as I can change to the already open browser and click the refresh 
button, a second or two (after waiting for the update to complete of 
course).

I tried an experiment. I changed a large file on the local copy of one 
of my sites. I then started an update, involving only that file, that 
took about a minute. I immediately tried to display the updating page on 
my browser (FF). There was no display (just a white browser window) and 
the progress bar did not even start to advance. I watched the progress 
of the update and as soon as it completed the browser progress bar 
started to advance and completed. However, I got an error message that 
said the file was "broken" and couldn't be repaired. The page I was 
downloading was a pdf file and the error was from Adobe Reader. Perhaps 
this wasn't the best file type to select but it was one of the larger 
files I had available.

One thing was obvious, the server appears to lock the file once it 
detects it's being updated because the delay before the page started 
downloading was unmistakable. This would imply that there is no caching 
at the hosting site as I speculated in an earlier message. I belong to 
another group that has some discussion of hosting services and I've 
never heard of this "delay" issue before. I'm reasonably certain that 
most hosts don't work this way or there would have been a general 
uprising among developers long ago.

It's my understanding that the browser (FF at least) checks the 
date/time stamp on a page _every time._  If it's the same as the version 
in your cache it'll display the cache version to save downloading the 
page. If it's different it'll get the new page. But......

I just looked at the IE6 Options and it has these possibilities:

Check for newer version of stored pages:
Every visit to the page
Every time you start Internet Explorer
Automatically (what ever that means)
Never

It seems that Never might get you in trouble but Every time you start 
Internet Explorer might be what you're seeing??? Since you're using 
Maxthon maybe these options are "confused" somehow.

If that's not the problem maybe there's some way you could trace the 
route of your browser request to the server. I'm not up on these 
techniques but there are plenty of trace route programs available and 
there must be a way to specify which port on the server you want to send 
the request to. You want to test to the http port not the ftp port. Or 
trace to both and see if there are any differences??? Perhaps some group 
members that know more than I do about this can offer more help.

I was also wondering if this is a new behavior. Have you been updating 
your site for some time without having this problem? Or has this problem 
always been present?

I think this could be a good candidate for problem of the year, and it's 
only February.

H Davis

Don Penlington wrote:
> I think I've figured out what's happening, thanks to your many and varied 
> suggestions.
>
> When any web page is called up by a viewer, it remains in the host's cache 
> for a certain number of hours before it gets deleted.
>
> This would, I think, explain why the ftp software shows the newly-loaded 
> page ("View source") on the host but the browser is downloading the old 
> page from the cache. If I "view source" on the downloaded page in browser, 
> it shows the old coding.
>
> I tried Will's idea of deleting the page before re-loading it, but that 
> didn't make any difference. Nor would I expect it to if my explanation is 
> correct, as the old page would remain in the server cache.
>
> Does that sound logical?
>
> It's a darned nuisance when you're in the middle of refreshing a website, 
> as you can't check your results until next day. But it does result in 
> almost instant downloads of each page.
>
> Do all web hosts work that way I wonder?
>
> Another possibility I guess is that my ISP caches all web pages for a 
> certain time, though that wouldn't make a lot of sense as it would 
> effectively disable the refresh button.
>
> Don Penlington
>  From the Beach at Surfers Paradise in sunny Queensland.
> Computer tutorials, local scenery,  and other things at my website:
> http://users.tpg.com.au/deepend/index1.html
>
>
> _______________________________________________
> Tech Support Guy Mailing List
> http://www.tsgserver.com/list/
>
>   

-- 
H Davis   hdavis1 at gmail.com