Remove outdated IE5-mac hacks for build_css_cache regex

Just as an FYI, I believe you can also use an "@media" block to hide styles from IE5 mac. eg:

@media all {
  #nav_main a { float: none; } /* not mac ie 5 */
}

Has the bonus of not relying on a parsing quirk to do its job and should therefore be more robust in the face of different CSS compression techniques.

Replacing the commented backslash hack with an @media all { ... } block should fix this issue without any side effects (that I know about). That's the form we use in our standard stylesheets.

IMO, the solution isn't to make the compressor more complex to handle badly formed code, it's to remove the badly formed code from Drupal in the first place. Let's either use @media for that code block or simply remove the IE-Mac hiding entirely. (I mean really, that's like supporting Netscape 3.)

I'm all for removing the "IE/Mac"-hack.. Any selfrepecting person on a Mac brows' the web with Safri (free from Apple), Camina or Firefox..

Let's face it, IE for Mac is dead since Apple released Safari and Mozilla released Firefox for Mac. Let's remove this ugly piece of code.

Committed. :)

This patch is silly. By removing the two comments, but leaving everything else intact, the .clear-block { display: inline-block; } is now always overridden by the later .clear-block { display: block; } style. It's dead CSS and it takes a whopping 5 seconds of looking at defaults.css to discover this.

The excuse to "leave the position-is-everything hack intact" is silly and just an excuse to not dig deeper.

I've rolled back the patch so we can have a better look at it.

Unfortunately I believe that the display { inline-block; } is still required (or at least something else needs to be added) for IE 7 support.

IE 7 doesn't trigger the hasLayout property (responsible for it's float clearing behaviour) when there is a height set as older IE's used to do. So the current 'height: 1%;' will only work for IE 6 and below.

What DOES trigger hasLayout in IE 7 though is setting 'display' to 'inline-block'. Another aspect of hasLayout is that once it's set on an element it can't be removed, so despite the 'display' property later being changed to 'block', the hasLayout triggering side effect remains.

More info on this is here if you're interested: http://www.tanfa.co.uk/archives/show.asp?var=300

Another alternative is to put 'zoom: 1;' directly after the 'height: 1%;'. The 'zoom' property is another hasLayout trigger that works in IE 6 & 7. If we don't care about Win IE < 6 then we could also ditch the 'height: 1%;' altogether.

Only necessary bug fixes will go into Drupal 5.x to avoid potentially breaking themes.

If we're not going to port this to D5.x branch, then I believe we should use the @media all to fix the bug for D6.x and get rid of as much CSS hacks as possible. If removing this CSS hack introduces bugs for IE7, then we can always add this to the special ie7-fix.css stylesheet that ships with core.

Frankly, some of it was destroying our D5 themes, and I can't bear to see this code remain in D6. Experienced designers almost always know what hacks to apply to get their themes working correctly.

Here's a new patch which works for 100%.
It drops IE:Mac support, and fixes up the hasLayout problems in IE7.

Please review and apply...

Forgot the patch... (bummer)

Patch still applied against CVS HEAD.

The code looks okay, but it should be run by the W3C CSS validator as well.

See: #335122: Test clean HEAD after every commit and http://pastebin.ca/1258476

Tagging. There are duplicate/related issues.

Lets drop support for IE-Mac in D7. Is it actually used?

IE6 doesn't *support* backslash hacks!

hmm, personally I wouldn't fiddle with the clearfix stuff, it's been heavily tested and backwards compatibility has no downsides at all in this case

also, doesn't IE7 need that inline-block to block stuff?

Drupal's clearfix is based upon http://www.positioniseverything.net/easyclearing.html

display: inline-block; is for IE for Mac only which has a market share worldwide of like 1E-10 percent. It's total nonsense to support that browser.

The exception in the regexp for removing comments is only there for the IE-Mac backslash hack (http://www.sam-i-am.com/work/sandbox/css/mac_ie5_hack.html).

@casey: see #17, the inline-block isn't just for IE5, it also triggers hasLayout for IE7, even if it's overridden immediately after by the "display: block" declaration. If we remove it, we need to include a zoom:1 or something that triggers hasLayout to ensure that this still works for IE7 as the "height: 1%" line which does this for IE 5 & 6 doesn't work for IE7.

Just found this issue.

GOD, YES!

IE5/mac is long dead and having support for it in D7 is just embarrassing. It also causing us a great deal of pain. See #444228: Optimize CSS option causes php cgi to segfault in pcre function "match"

I don't see us removing the clearfix class from D7 (I hope from D8 though). So we need an updated version of clearfix without the IE5/mac hack.

seutje and ejort are however right on my patch; it's not correct.

http://www.webtoolkit.info/css-clearfix.html gives an alternate clearfix which might the easiest solution for us.

Over in #444228: Optimize CSS option causes php cgi to segfault in pcre function "match", I noticed this comment and am posting my response here since its more appropriate.

If we follow this path, we have to place a warning at the "Performance page" bellow the "Enable CSS optimization" about hacks not supported.

I'd like to state for the record, that I disagree with that assertion. If IE for Mac is not supported, we just list that on the Drupal project's requirements page. We don't need to litter the UI with warnings about a long dead obscure browser not working with a particular feature.

If we do this, we can solve #444228: Optimize CSS option causes php cgi to segfault in pcre function "match" by removing support for IE-mac hacks (nobody uses that browser and we never tested it anyway).

Reviewed the proposed changes very carefully. I've been using clearfix successfully for 6 years, so changes always need to be scrutinized. I spent about an hour making sure all the CSS is necessary. The IE7 "addition" is necessary because the old MacIE5 "display: inline-block" rule also triggered hasLayout in IE7. Since we're removing the IE5 bit we need the IE7 line. Even still, this is less code. and will help out with #444228: Optimize CSS option causes php cgi to segfault in pcre function "match".

Thanks, casey!

Still applies + #444228: Optimize CSS option causes php cgi to segfault in pcre function "match" landed.

Actually, this might be critical, now that we removed the support for MacIE5 hacks from the CSS aggregator :)

Committed to CVS HEAD. Thanks.

Is there any value in backporting this? I mean, IE-Mac probably wouldn't even work on a modern OSX machine anyway and has been discontinued for over five years (as compared to IE6 which is still officially supported).