Should Netscape have released Version 6 already?
11 December 2000
Yes, it’s another CodeBitch column on Netscape 6. I’m sorry about that, but they did insist on releasing PR3 right after I did my piece on PR2, so now I’ve been cornered into keeping my words up to date.
In my earlier columns, I complained about the incomplete CSS1 support in the preview
releases, including some hideous regression bugs – things that were
fixed in earlier versions only to become broken again. I’m sorry to
report that only a couple of these appear to be fixed in the final release
(you can check this for yourself by following the links in my column on the subject). The problem in PR3
with :firstletter formatting disappearing intermittently seems
to have been fixed, as have the problems with hover text that were in PR3
but not PR2. As far as I can tell, however, all of the other problems I
identified remain.
But were these glitches enough to warrant delaying the release of the browser, as some have argued? I’m not so sure.
Naturally, I’m disappointed that Netscape 6 didn’t live up to its promises of standards compliance. For all intents and purposes, IE5 for the Mac is comparable to Netscape 6 for the number of glitches in its standards compliance (the same certainly can’t be said for IE5 or 5.5 for Windows). Netscape 6 does have better CSS1 support, but IE5/Mac only ever claimed CSS1 compliance anyway, and lived up to that reasonably well. IE also crashes much less often. In any case, Netscape’s glitches in standards compliance are pretty minor things – things that most Web authors aren’t using anyway, because they never worked in Netscape 4.x.
What should have caused Netscape’s management to delay a final release was the appalling speed and instability of the current release. The Web is full of sites complaining about this. I have really, really tried to give Netscape 6 a decent workout since it was released. But it won’t run without crashing for long enough. I’ve encountered some memory leaks that make the Finder complain about low memory, and since my usual home machine only has 64 MB of RAM, I can hardly run anything else at the same time, which is not how I like to work. The default memory partition is only about 20 MB, but just opening it on my machine takes a 24 MB slice of RAM, opening a single simple page off my hard drive takes it to 25.7 MB, and a couple of other windows will add another megabyte without even trying. Ten minutes of use will take it to 33 MB. Java applets that work in IE will cause Netscape 6 to quit (and yes, I have the latest version of Macintosh Runtime for Java). This is not the sort of experience that will encourage current users of Netscape 4.x to upgrade.
Do as I say, not as I do
Despite my criticisms, I really do want current Netscape users to upgrade. If every current Netscape user upgraded, we would finally be in a position to get rid of the ridiculous hacks that everybody has to go through to use CSS on their pages – or pretty much anything that’s been introduced to the Web since 1995. Netscape 4.x was an abomination that should have been withdrawn from commercial release as soon as Version 6 went final. The fact that you can still easily download the latest point release of Communicator from the same part of Netscape’s site as Version 6 is disturbing. Communicator should have been moved to the part of the site where old point releases are located, so that the only people who will download them are Web authors trying to work out just which point releases of Navigator 4.0x crash on their HTML 4-valid, syntactically correct pages. (Am I bitter? Oh yes, indeed! These browsers crash and have outdated digital certificates – so what are people waiting for?)
Pointing the finger where it’s due
A lot of the complaints about Netscape 6 relate to the way it chokes on
certain Websites. Problem is, half the time it’s the fault of bad
HTML markup. Web authors say they want standards compliance, but there are
two halves to that particular equation. Browser manufacturers can support
standards, but Web authors have to stick to
them as well. For example, one of the sites that MacEdition staff
visit regularly (a conferencing site we use for scheduling) is completely
broken in Netscape 6, with black text on a black background. This
isn’t really the browser’s fault. It turns out that part of
the page opens a TABLE tag before it closes the previous DIV
tag, so that the markup goes: <div>blah blah blah<table
...></div><tr>.... No wonder Netscape chokes. What
were the developers of this site thinking?
I can understand missing subtle parts of the standards like &s in URLs or comment syntax, or even the common P-inside-FONT boo-boo. But this is just bad nesting, a real kiddie error.
I don’t blame Netscape 6 for choking on these sites, and neither should other Web designers. A browser’s ability to render valid HTML is a very different thing from its ability to render bad HTML. The standards compliance that Web designers have been campaigning for requires browsers to stop trying to make up for bad markup. Moreover, some test suites, such as David Baron’s and Ian Hickson’s pages, penalise browsers for not choking on bad HTML.
Small compensations
At least now that the product is out, we no longer have to bow and scrape to the open sourcers and say, yes, you wonderful Bearded High Priests are going to give us something truly magnificent and save us from the clutches of Microsoft. Crap code is crap code, whether from Redmond or the crowing GNUsters. No product is perfect, as Netscape’s release notes remind us, and now that it’s released we can call the compliance as we see it. Netscape no longer has the moral high ground on these issues, now that it can be judged on deeds, not words. In the scheme of things, it’s made a good fist of standards compliance. It’s just a shame about the rest of the browser’s performance.