schepp / css-js-booster Goto Github PK

hi
is it possible to CSS image path "/booster/booster/../examples/example1/css/../images/header.jpg" is simplify like "/booster/examples/example1/images/header.jpg"

If we use only config file parameters then there's no need to check for doc root inclusion i guess...

Since we're planning to pass parameter by config files, why not now have calls to css and js files like this :

/CSS-JS-Booster/booster/css.md5.timestamp.datauri.css
/CSS-JS-Booster/booster/css.md5.timestamp.mhtml.css
/CSS-JS-Booster/booster/js.md5.timestamp.js

With help of htaccess of course it could also means that the web server could direclty read the files instead of calling a php script.

Apache can do rewrite rules to css.php or js.php only based on the fact that "does the file reall exists?" => yes => give it to me / no => call css.php or js.php based on file request

It would be safer i think.

I also know this is huge work but will definitely give css-js-booster a big advantage in the dynamic css js generation "world" :)

Method Booster::css_split() does a split without taking opened @media-rules into account. So the resulting CSS contains errors which could lead to incorrect CSS rendering.

Just let the system do his magic.

If the webserver can create the file, he'll be able to modify it, no problem with this.

Hi, i'm using css js booster but ran on a bug :

on http://mywebsite.com homepage, with http://mywebsite.com/CSS-JS-Booster present, all is OK.

When outputing ->css_markup(); the href of the css stylesheet is "CSS-JS-Booster/booster/.." then i go to http://mywebsite.com/page1/ wich is in fact redirected to http://mywebsite.com/index.php with a rewrite rule, this is a fake directory page.

In this cas, the href is still "CSS-JS-Booster/../.." but as we're on "http://mywebsite.com/page1/" browser will try to get "http://mywebsite.com/page1/CSS-JS-Booster" wich is not a valid directory...

How can i force the css js booster path if i know it ?

I think you should always write the full path to the css js booster, not a relative one

Hi,
I can't seem to get any CSS files to work in my site which has an MVC type structure. I suspect there's conflicting re-writes going on.

Javascript files work with some success i.e Jquery refuses to play nicely no matter what and isn't available to the libraries that need it, but a lot of other files do work fine.

Great bit of coding though!

...instead of compressing on the fly all the time.

Test text

With CSS-JS-Booster calls it is possible to access webroot and non webroot files like /etc/passwd or any php file that contains passwords and such.

In my fork i double check the file extension and file mime type (though javascript and css files doesn't have specific mime types...)

But i think the best would be to have configuration files (in json format) that stores the file groups, passing the files group to uri is very dangerous as said by zoompf http://zoompf.com/blog/2009/12/the-challenge-of-dynamically-generating-static-content.

And he's right because in default version of css js booster we can access the whole filesystem !

Dunno why but css hosted minifier is not working anymore with 1.6.x

If the CSS contains multiple references to the same image, CSS-JS-Booster currently embeds the image for each occurance. This can easily lead to a pretty large CSS file (or many chunks of it).

MHTML does not fall for this. but with data-uris I don't see a viable solution to reducing the overhead. The only way out is to skip images with multiple references. If the WebDev still wishes to make use of data-uris, he must modify his CSS accordingly.

I downloaded the current 1.6.3 version and installed on our box (Ubuntu 9.10). I get an error at first requiring to create the booster-cache folder. I did. But the content of the .css file generated is just "css" and the javascript seem missing nearly all the code.

    // Checking if Apache runs with mod_rewrite, @dded by xod: When PHP is run as CGI Script this throws a fatal error, so I added the function_exists checking for the whole thing...
    if ( function_exists( 'apache_get_modules' ) ) {
        $apache_modules = apache_get_modules();
        if(in_array('mod_rewrite',$apache_modules)) $this->mod_rewrite = TRUE;
    }

just recognized that and changed it a bit, now it works =), maybe you want to implement this?

For Squid etc

I am the author of this question: http://stackoverflow.com/questions/3501295/packing-caching-js-and-css-in-php-that-differentiate-between-development-and-pr

I read a lot of articles and solution around. CSS-JS-Booster looks like the most close to what I need.

In order to get even better optimization read the question in stackoverflow

This is pretty much what you are missing:

1- Option to turn on/off debug mode.
2- To improve development of CSS styles support .less files compilation and use of LESS.js for debug and compiled version for not debug mode.
3 - Instead on using JSMin, use Google Closure Compiler that make a better job with Javascript.

When calling cssmarkup/jsmarkup, cachedir is still "booster_cache"

Since getwcd command on my system reports /home/http/nameofwebsite/www, error check pass and reports "booster_cache" is not a dir (right, it this in /home/http/nameofwebsite/www/CSS-JS-Booster/booster/)

But the next functions works because they call setcachedir.

Also how setting the cachedir is hard, i do not understand (trust me i'm lost, this is not free criticizing of your work).

Using booster_cache = realpath(dirname(FILE)).'/booster_cache' should do the work ?

Hi Christian,
ich habe es gerade bereits bei Peter im Blog gepostet, aber will nochmal den offziellen Weg einschlagen. :-)

Nachdem ich den Booster als Plugin bei mir installiert und aktiviert habe haut es beim Aufruf der Seite oben den String "--> --> " raus ...
Seite: http://www.schnellkochtopf-rezept.de

Ich habe das Plugin erstmal wieder deaktiviert. Kann es aber durchaus wieder einschalten, wenn du es brauchst.

Ich hoffe, du kannst helfen ... :-)

Sonnige Grüße vom Bodensee
Jörg

Case :
i update my css and js files, push the updates to server then remove the files in booster_cache.

If a client with a full cache (means that he already visited my website and have the file in cache because of max-age directives) goes to the website then he'll not get the new js and css files because he'll not ask for them as long as the number of seconds in max age has past.

What i ask is to have a versioning system like this : css.php?...&lastmodified=timestamp then the ressource will have the timestamp in his URI and if the timestamp changes the client will ask the ressource again.

Query parameters are dangerous !

what happens when if i change "cachedir", when i change "hosted_minifier" etc...

Security is high concern and i do not want potential attacker to be able to generate a lot of files freely on the server.

Hi! I'm currently working on integrating css-js-booster into an existing website that uses .css files scattered throughout its directory tree. I'm assigning an array containing paths to the .css files relative to the booster_inc.php to $booster->css_source. Also, I'm developing my website at http://localhost/~username/foobar, so I am setting $booster->base_offset and $booster->document_root, too.

Everything seems to work out fine, except for a wrong path calculation in css_datauri_cleanup(). In booster_inc.php:1144, $booster_path is calculated using the document_root object-property, which has never been set for this instance of the Booster class that executes this function. css_datauri_cleanup() is executed only on the instance which was constructed in booster_css.php, and naturally, that instance knows nothing of the
document_root I used when parameterising $booster in my website's files.

However, I don't think it's actually necessary to recalculate $booster_path. In my opinion, it should be enough to use the $dir paramter that gets passed in with a usable relative path anyway? So, in line 1169, instead of

$replace = $treffer[1][$i].$booster_path.'/';

it should be enough to say

$replace = $treffer[1][$i];

At least, that seems to work for me. What are your thoughts on this? Am I missing something?

Regards,
please keep up the good work!,

Fritz.

ps: same file, unrelated issue, line 1170: I guess it should read if (...) ... = $css_stringbase instead of ... $css_stringmode.

When using query strings for ressources, it disable shared caching

http://www.stevesouders.com/blog/2008/08/23/revving-filenames-dont-use-querystring/

What i want is to have css and js files called this way : css/dir=hello1,salut2,hi&blabla=1&blabli=0 (i think & are allowed since minify does it like that)

Or why not trying json objects in query string, easier to parse ?

css/{"dir":["hello","hello2","hello3"],"blabla":0}

a bit weird but very easy to undertand

"Resource interpreted as image but transferred with MIME type image/jpg." <- who's fault is that? Chrome's or Booster's?

On one of my websites the eot file of a referenced webfont was included as data uri by the booster. As a result IE8 didn’t display the webfont. I didn’t test this with IE7 and IE6, but I think it should be a similar result.

I'm talking about this : https://github.com/Schepp/CSS-JS-Booster/issues/closed#issue/15

The "lock file" when dealing with a large number of accesses and trying to generate a newly cached file.

The url of the temporary merged file is the same of the final completely-merged-minified file. That means that a CDN will try to cache the temporary file and serve it to all clients.

I hope I'm clear! :)

All assets should have last modified headers

Problem :
You have a 404 page with a very large background (but < 24ko), jpg style
You have 2 css files : core + template.
You use css js booster to merge, minify and data uri your code. Then the large background is data uried and you have a BIG css file.

Solution :
When calling css markup, if we could provide a ignore file with on each line some characters like :

404
big
big_image

Then when css js booster parses the CSS file, when he find an image he wants to data uri, it has to check if the image uri does not contain one of the keywords of the ignore file

This could be an ignore file but an array or a string can do the work too.

I know this sounds a weird idea and that could be solved by making a css file only for the 404 page, then you lost the merging benefits.

What do you think ?

I have tried loading your plug-in several times from source by downloading the WP specific plugin as well as V1.8 from here. I am scratching my head as to why this is not loading. I have attached my source code below.

Note: I have used the sample file on the same server and it works perfectly. I am excited to use the plug-in I hope you can give me some insight on why it may not be working.

Html source: http://pastie.org/1273362
Website: http://bit.ly/96q3ig

I am not utilizing any other plugins besides WP Importer.

Thanks in advance,

J Nolte

Hi again :)

all is in the title,

i got one call with 2 folders then on unique file for css source
if i update the unique file i do not get the cache to update

i got a second call with only one css file for css source
if i update the file, the cache get updated

i think getfilestime only parse first folder .. just an idea hope it helps

In my (smarty based) projects I usually include a small plugin to remove whitespace and comments from HTML as well. (yes, ignoring <script|blockquote|pre|..>, yes, without destroying conditional comments.) Where apllicable (and necessary) inline CSS and JS are minified to.

Most effort is put into optimizing CSS and JS. But once they've reached the client (and were fitted with proper HTTP-headers for caching), they are of no further concern. What about the HTML itself?

We've basically got 2 types of HTML. Static, (public-cachable) HTML and user-specific (private-cachable) HTML. The static parts behave just like CSS and JS once optimized and cached. But especially those not-public-cachable or not at all cachable parts benefit from deflating a lot.

Yes, this will not speed up page loads as drastically as reducing the HTTP requests. But over time you'll see your traffic plummeting. We did.

(no gzipping of WOFF)

Problem : on a high traffic website, there can be hundreds request to pages in one or two seconds.
If the generation takes more than 2 seconds then you'll launch a lot of css js booster file generation for nothing and the machine could crash (will crash)

Solution :
generating a lock file before doing the generation, when it's done delete the lock file
If the user ask for the files and the lock file is here, give it raw css and raw js with expires:now and max age 0
then the next page when the generating is done he'll get clean minified cached css and js

I know we already discussed the problem by DM but this is a reminder for both of us :)

They break the code.

Hi Christian,

ich habe nun mit Mühen meine Seite http://www.schnellkochtopf-rezept.de wieder valide und bin dabei dahintergekommen, dass meine Probleme zum Teil auf CSS-JS-Booster zurückzuführen sind. Ohne das aktivierte Plugin validiert die Seite. Mit Plugin nicht.
Ich habe das Plugin mal aktiviert gelassen. Du kannst also mit dem W3C-Validator den Fehler sehen. Irgendwie beißen sich Comments mit dem BadBehavior-Code. Ist dein Plugin aktiviert und BadBehavior deaktiviert, validiert die Seite auch wieder. :-)

Viele Grüße vom verregneten Bodensee
Jörg

What behavior i expected :
i'm using hosted minifier (no debug mode), i have only one js files made of multiple folders/files.
If i modify one file : the nocache argument is changed (great) and the browser redo a request for the JS.

But the JS is automatically taken form the cache (altough on file have changed)

here are my parameters :
$js_files = 'folder/folder,folder2,folder2';
$booster->js_hosted_minifier = TRUE;
$booster->js_totalparts = 1;
$booster->js_source = $js_files;
$booster->booster_cachedir_autocleanup = FALSE;
$booster->js_executionmode = "defer";

it should do the same for css files

thanks

...if they reside on the same machine as Booster, e.g.:
"/booster/booster_js.php/dir=http://domain.com/script.js"

When an @import or file-resource is within comments and therefore disabled, Booster should not process them, e.g.:

/* @import url(file.css); */

Especially with outcommented @import this might lead to comment nesting errors (imagine there would be a single */ within file.css).

I found out that the CSS parser makes a mistake if you use the @page decleration inside a @media print.

I hope you can fix it ;)

Could the 24KB limit be raised if the page lazy-loads data-uris?

There's cases where the ressource that says "max age xxx" will get cached by a strong cache system (like varnish), then there will be only one cache for all.

Then IE-6-7 can have data uris or All others can have mhtml (based on who generated the cache first).

We'll have to use conditionnal comments to get this working.

Also do not forget to add empty conditionnal comment, explained here http://www.phpied.com/conditional-comments-block-downloads/.

Hi Schepp,

I wanted to deactivate the CSS-JS-Booster plugin because CSS updates didn't take effect. After I clicked on "Deactivate" it started to load and after a few seconds a Wordpress "Service" Error occurred. This happend four times.

After that I removed the folder. In WP Plugin overview the plugin is gone but the css still seems to be in a boostercache file.

Can you help me?

example: http://quaeng.de/bilder-u4-tunneltour-u-bahn-tunnel-ueberseequartier-hafencity-jungfernstieg-hvv-verkehr-9212/

Regards

Patrick

Hi, when I try to activate the plugin I get this error:

Parse error: syntax error, unexpected T_STRING, expecting T_OLD_FUNCTION or T_FUNCTION or T_VAR or '}' in /home/graphi9/public_html/wellcore/wp-content/plugins/css-js-booster/booster_inc.php on line 88

Just a very small modification...line 1798

        if ( $this->css_title != '' ) {     
            '" title="'.htmlentities($this->css_title,ENT_QUOTES).
        } else {  }