Raul P.

@rottame I have it running on PHP 7.4 and 8.0 without issues, but each setup can be different. You can try to disable features one by one and see if it fixes the issue.

That error is for the html minification, so you could turn it off and see.
But it’s strange none of the other uploads work, as they are mostly a rollback to 3.3.1

Either way, if updating the docker image works, then all good.

Thanks

just to confirm, @rottame you cleared the cache on w3tc and the timestamp changed on the source, but still blank, correct?

The file I uploaded makes use of w3tc buffer, so there should be no conflicts with it.
Are there are error logs on the server?

If the previous upload doesn’t work, please also try this:

# changed buffer to 3.3.1 location only
b) https://dl.dropboxusercontent.com/s/7gsbo3207dcaz57/bVh43pb8p6.zip

or

# changed buffer to 3.3.1 location + 3.3.1 php minify version
c) https://dl.dropboxusercontent.com/s/hghmzzi5op5p621/NMWBPiJupL.zip

I did not see any issue with the minification, but I also changed how the buffer works on 3.3.3 and it’s a common issue for plugins to conflict when several of them use buffers.

I have improved compatibility for wp super cache and a few other cache plugins.
Please try this update, and if it works fine for you, I will push it as a release update.
a) https://dl.dropboxusercontent.com/s/c7kzymph8h7ugpn/5DlKGeJM3g.zip

@xxxlesy @rottame

Thanks

That is being output instead of css code?
Looks like a mix of javascript and css, so I am not sure how could that happen.
Let me do some tests and get back to you later on this.

PHP Minify was updated on 3.3.3 version, so likely there is some incompatibility.
Can you provide your url so I can take a look at the source html?

Usually when a page is cut off, it means that the HTML is not valid and you either forgot a closing or opening html tag. This is because FVM parse the structure DOM and if it’s broken, it also breaks.

You can ignore most errors, but take a look for possible missing html closing or opening tags: https://validator.w3.org/

If the markup is valid, then send me the url so I can simulate on my end processing that page.

The $_SERVER['REQUEST_METHOD'] is a server variable set by the server, and the only time it does not exist is when you call it via command line on the server or if your PHP is misconfigured somewhere.

Maybe you have a manual cronjob set by your server, or perhaps some plugin or service is running the site directly on the server without HTTPs… but it’s just an explanation and it doesn’t matter.

I will add a change to check if this variable exists before actually calling it, so it will fix the error log regardless.

Thanks

There was a buffer error, but please update now and it should work. Thank you for reporting.

As for the REQUEST_METHOD part, if it still shows please check if are not running it via cli, instead of via https.

Please update now and it should work. Thank you for reporting

Thanks for reporting, there was a bug now fixed. Please update.

This will be up shortly, thanks!

This will be up shortly, thanks!

The query generation on FVM did not change between past and latest version. FVM does not process a page when there is a query string, except for some situations where those are recognized as not changing the content (utm analytics for example).

From what I see, w3tc preload for mobile is being worked on, but not available yet, so I understand why you made your own.
https://www.ads-software.com/support/topic/mobile-cache-preload/

The issue is with the settings probably, regarding query strings. There are usually two ways to cache query strings.

One type will ignore certain query strings and cache the page as if they don’t exist. Another method caches the query strings separately, as individual different pages.

Unless those are tracking query strings that don’t affect the displayed content, they should be cached individually.

According to this link https://wordpress.stackexchange.com/questions/264963/w3-total-cache-site-with-query-strings when you add the query strings, they become ignored, so they effectively become the same page, so if you use query strings to show different content, probably some are broken when cache is enabled.

Can you share your bash script?
How you can do it for example, is to read the xml sitemap and then curl each url with a mobile useragent and again with a desktop useragent, then disable preloading on w3tc if you wish.

Visiting each url twice, one with a desktop and another with a mobile useragent should cache the pages correctly, as there are no query strings on the sitemap (probably).

FVM does not process pages when most query strings exist, because there are plugins and anp endpoints, using non standard url’s. Optimizing those pages, would break the api in most cases.

Thanks, I also find it weird, because that means w3tcache is creating a cache file when FVM thinks that it should not be creating the min files.

Guessing here but that “could” happen if the page is being cache expired and then some other visit triggered the cache regeneration differently, when FVM thinks it should not optimize the page.

For example, if your site is example.com and it responds to example.com, FVM would generate the file. But if your site opens when you access the IP address directly, or if you had another virtual host pointing to that IP before that has been removed from that server (even if it did not belong to you), that could cause some bot or someone to open the url or ip, not matching your hostname.

FVM will not generate the files or include them, if the hostname doesn’t match the actual site domain.

Do you have some preloading feature going on (crawling to prewarm caching) active or customized somewhere?

Is your w3tc setup to ignore certain query strings, and creating standard caching files on certain queries? FVM also does not generate css/js files for most query strings, so if a page was cached during a query string request, FVM would not process it.

If you can export the w3tc cache settings and send them to me to info (at) fastvelocity.com I can try to replicate or make some tests.

w3 total has it’s own minification feature, so you can actually use it directly without fvm.

That being said, unoptimized pages will show up if you are logged in, or have some cookie setup. When that happens, please check with incognito mode on google chrome if it’s still unoptimized, or if it’s specific to your session.

Also if you want to use fvm with w3t you need to disable everything related to html, css and js optimization on w3t, else they will conflict.

Also if your server or hosting already has page caching, that could happen, as it may cache the page before it can be generated.

• This reply was modified 2 years ago by Raul P..

sorry about that, fixed now.