Newer browsers less strict?

[Anonymously Famous]

Some background:
I work in a translation company. We were asked by a client to take a look at the Spanish version of their website, for linguistic errors and such, and get back to them.

What we found was this: Some of the words with accented characters such as "más" (more) displayed as "m&aacutes" in Firefox 3.6. They got back to the client, who couldn't see the problem. It turns out that in Firefox 5 and Chrome, it displays the word as intended. We later advised the client to either go through and put the missing semicolon on the entity or just use the character itself.

So my question is this: Is this sort of smart correcting good or bad? I know that browsers have to be somewhat robust, that people don't always write well-formed HTML, but some things are clearly errors. What sorts of things should the browser pick up on?

[Xanthir]

What's important, more than anything else, is that every browser corrects things in the same way. These days, modulo an occasional bug (which you should report), they do. So don't worry about the details too much, just be happy that HTML5 reverse-engineered everybody and set down a single true way to do all of this.

[Anonymously Famous]

What's important, more than anything else, is that every browser corrects things in the same way.

This would be true, but they don't. The browser that it didn't work on was Firefox 3.6. A month ago or so, that was the most up-to-date browser there was.

[Xanthir]

Reread my statement. I'm talking about modern browsers. Firefox 3.6 does not use the HTML5 parser by default (it's there, but you have to turn it on with a pref).

Almost every behavior in the HTML5 parser is there because at least one browser (usually IE) was already doing it. (Occasionally, every browser did something different and they were all stupid, so it wasn't too bad to make up a new behavior that was sane.)

[Anonymously Famous]

In an ideal world, you may be right. I know a lot of people who do not use "modern" browsers.

And I'm fine with HTML5. The missing semicolon is not HTML5 compliant, though. It's just incorrect HTML. I'm not just talking about the web entity, either. I'm talking about missing end tags, etc. that parsers have to "guess" at. Is it, or has it been, really beneficial in the long run to be forgiving of these errors?

[Xanthir]

It doesn't matter if it's beneficial in the long run. It's necessary to correctly display large amounts of content on the web. Every once in a while we find some detail that isn't necessary for web compat and fix it to be sane, but otherwise all the weirdness of modern browsers is intentional and we're stuck with it.

That said, definitely tell you clients to stop fucking around and just input the entities correctly, or use actual unicode characters. What they're doing is indeed invalid, even if HTML specifies exactly how browsers are supposed to handle it.

[Goplat]

It doesn't matter if it's beneficial in the long run. It's necessary to correctly display large amounts of content on the web.

Having a bunch of "&amp"s everywhere doesn't prevent people from using a site, it just makes the site look as unprofessional as it actually is. When a page is doing something obviously wrong, like using unterminated character entities, then it should look wrong.

[Xanthir]

Your disapproval means little to the browser vendors, who are motivated by market forces to display pages as close to the author's intent as possible.

[Yakk]

Be strict in what you produce and generous in what you accept.

[Goplat]

Be strict in what you produce and generous in what you accept.

I'd say this principle is a failure in practice. The few people who write browsers followed this principle, but each had their own idea of how to be generous. The masses of website authors knew nothing of the HTML standard or this principle, so each just did whatever works in his own preferred browser. The inevitable result was lots of sites that look right only in some browsers and not in others.

[Yakk]

Yep. So now html5 specifies exactly how to be generous.