Blogger, you ignorant slut

Finally got sick of waiting for a fix for the way Blog*Spot currently works with archives turned off (Blogger then publishes <$BlogItemArchiveFilename$> as “index.html”, but during the B*S upgrade/reorganization a while back they stopped redirecting index.html to the actual file in /homes/thesubdomain.html, so if you use one of the templates Blogger provides and turn archives off, rather than having permalinks that point to the post in the main page, which is what I wanted for profaq.blogspot.com, you instead have permalinks that are 404ed).

<update>Since fixed.</update>

So, I took my brand spanking new copy of Phoenix 0.5 over to Blogger.com to edit my template. Oops, whatever nasty combination of invalid HTML and mixed-up encoding from assembling the page from multiple servers it is that causes Mozilla-based browsers to eat the attributes in every Blogger archive related tag in the template textarea, turning <a href=”<$BlogItemNumber$>”> into <a> <update> the odd behavior of Gecko-based browsers on encountering an unescaped </noscript> tag in the content of a <textarea>, where they then strip off the attributes from any following tags </update>, is still there.

Not a problem, I’ll just fire up w.bloggar and edit the template there. Except, as I advised someone the other day, sometimes the paperclip that connects Blogger’s template server(s) to the various web servers comes loose, and to fix it you have to save changes on your template (without actually making any) on whatever web server you can find that knows about your template, to force it to copy the changes around to the other servers. In this case, the API server couldn’t find my template, but the Pro web server could. Yeah, the one where Phoenix eats the archive-related tag attributes.

Finally, I bit the bullet, started Internet Explorer for the first time in weeks, quickly saved the template before I became reinfected, and was then able to edit the template in w.bloggar. Ev and Steve (and probably Rudy, too)? You’re on my list now. bloggerAPI 2.0 better be slick, fun, and easy to make up for that.

24 Comments

Comment by ruzz #
2002-12-08 22:47:37

Wait.. this post wasnt about MT. I thought this had become a MT only blog.

dont let me down phil ;)

 
Comment by Phil Ringnalda #
2002-12-08 22:49:33

Oh, I think it’ll be pretty not-about-MT before long.

Though I thought it had become tiresomely all about RSS, not tiresomely all about MT.

 
Comment by Phil Ulrich #
2002-12-08 23:53:17

First off, I miss the RSS-related posts. You were my number once source for learning cool things to do with RSS and/or RDF.

Secondly, if you read between the lines, as I so often do, this post was about MT. ;)

And lastly. Ringnalda, you know as well as I do that the BloggerAPI 2.0 is going to be a bug-ridden POS despite the time it took to get it to public release. I fully expect no major changes over the BloggerAPI 1.0 except that it will probably support pro users. In which case, I will be making zero effort to support it with EspressoBlog, because I refuse to cough up the fee for Pro to test it.

Fugg THAT.

Well, that turned into a mini-rant of my own. Maybe I should go back to my own blog.

 
Comment by ruzz #
2002-12-09 00:01:09

I also read between the lines but was attempting to be funny. Guess I failed.

I don’t miss the RSS blogs, I think phil needs a RSS blog alone. I miss the stalking shannon blogs and anything else that might disprove my theory about Phil being a carefully assembled robot ;)

 
Comment by Phil Ringnalda #
2002-12-09 00:14:20

You would be surprised how little care went into the assembly.

And how can I stalk Shannon when she’s living with someone that her mom describes as not ”quite as scary as I remember him”?

 
Comment by nick #
2002-12-09 00:58:48

please let us all know in a month or two how many people are coming to your site looking for ignorant sluts. i’m already excited.

 
Comment by ruzz #
2002-12-09 01:17:07

A good stalker doesn’t let a menacing boyfriend slow him down!

Phil Phil Phil!

 
Comment by Ev. #
2002-12-09 11:51:59

I believe the attribute stripping is a bug in an older verion of gecko. What is the build date on that thing?

 
Comment by Phil Ringnalda #
2002-12-09 12:18:10

20021207.

And Phoenix builds off the trunk, so unless they messed with the forms widgets (not likely, not the kind of place they’ve been working), then it’s still in/regressed back in the trunk.

 
Comment by Ev. #
2002-12-09 13:12:36

Why does Mozilla not do that then? I’m using 20021111 right now, but it hasn’t done that for quite a while. It doesn’t seem like a server-side problem, as you suggest. But if anyone has a work-around, we’re all ears.

 
Comment by Phil Ringnalda #
2002-12-09 13:43:50

Two possibilities: either they regressed the trunk, or you just don’t happen to have a template with the right/wrong things that trigger it.

I thought I was on the trail of what does it, since I have two blogs with hosed templates, both of which have a doctype and a meta tag with a charset (3020551 and 3314035, the two faqs), and none of the others (some with doctypes, some without, none with charsets) were getting attributes stripped. But adding a doctype and charset to an unafflicted blog (3019985) didn’t mess up the template, or at least not yet. So maybe it does have something to do with passing it around through the template server(s) and web servers (if I’m guessing rightish about how that happens). We’ll see if that template goes south on another computer, or after a signout/session end, or just after a while.

Oddly enough, once I finally found the bug in Bugzilla, there has actually been some recent activity (you ought to point people who bitch about not getting an answer out of you ”when it’s been two hours since I emailled them” to some of the Mozilla bugs that have been going on for two or three years, with a month or two between comments): they say it’s your fault for not entity escaping HTML that you are putting into a textarea. I was pretty sure that was going to be the answer, and though I’m not quite sure I think they are actually right this time. The HTML spec says the model for textarea is #PCDATA as in Parsed Character Data, so you really shouldn’t be able to get away with putting unescaped markup in there. How expensive would it be to entity encode every template every time it’s requested? You shouldn’t actually have to do >, just converting every < to &lt; ought to keep Gecko from deciding to interpret it strangely. But then there’s the fact that they say you should convert every & to &amp;, at which point I start wondering whether you’ll break the current situation where only &amp;s in comments keep growing and growing, or whether the fix for that was something that contributed to the current problem.

In any case, while Googling I did run across this ouch. Ouch.

 
Comment by Phil Ringnalda #
2002-12-09 19:18:00

And the thing that triggers it (or at least the only thing in my templates that triggers it)? </noscript>. You know, whether or not it’s incorrect to put unencoded HTML in a textarea, tons of people do it, and I’d have to say that eating all the attributes on tags following a </noscript> is a pretty harsh penalty. Plus, it’s just weird. <noscript>? Not a problem. <script>? You bet. </noscript>? You infidel! I shall have your attributes for lunch!

 
Comment by Phil Ringnalda #
2002-12-10 00:32:51

And, Nick? It won’t take a month or two: now that Google’s back to over-ranking for freshness, I’m already number one for ignorant sluts.

 
Comment by Marcus #
2002-12-10 05:19:49

Mozilla bug 99467 seems to be covering the browser angle to the attribute-stripping problem… you might want to poke at that one too. It’s got a nicer CC list.

(157800 might even be a duplicate, depending.)

 
Comment by Phil Ringnalda #
2002-12-10 07:40:20

Maybe I just don’t have the right philosophy for Bugzilla, but I wouldn’t say that 157800 is a dup of 99467, and I also wouldn’t say that 156432, also marked as a dup of 99467 (with your grubby fingerprints on the verified, for shame!), which would have saved me the trouble of tearing my affected template apart tag by tag to find the problem, really is a dup.

In 99467, the problem is with a dynamically written <textarea> tag, with no contents except a link which gets its attributes eaten. In 156432 and 157800, the problem is with a </noscript> encountered in the content of a static textarea. They have the same symptom, and someone reading through the source might very well find that they have the same cause, but they might just as well not, only getting the same outcome by calling the same routine from two very different places. Maybe what I don’t get is that anyone working on Mozilla knows enough to go look at all the dups to be sure there isn’t better or completely different information in them, but if I was actually capable of doing anything with Mozilla’s source I’d expect to start out in a completely different place if I was working on a problem with a dynamically written <textarea> versus a static textarea containing a </noscript>, and I’d call 157800 a dup of 156432, but 156432 only ”possibly related” to 99467.

 
Comment by Marcus #
2002-12-11 03:16:40

You’re quite right – of course – but Bugzilla is often an illogical beast.

The way I see it is that there should be at least two bugs: an evangelism bug to convince Blogger to escape < and >, and a bug to stop Mozilla having a fit when someone doesn’t. At the moment we have 157800 and 99647. (In theory, 133518 should have been the evangelism bug that 157800 could have been duped to and, looking back, 156432 might still deserve to be separate…. but, ah well. Often it’s better to leave things alone when people have latched onto certain bugs.)

I duped 156432 to 99647 more because of general effect, than specific cause: ”This is wrong, but Mozilla shouldn’t strip attributes from tags when code is unescaped in textareas”. I assume that was the logic that Boris was working on too, as he brings up the dynamic / static distinction. Often when the symptoms and base cause are related so closely I’ll dupe them to make sure that people don’t get lost across Bugzilla, and that all the info is being brought together in one place. Sometimes it works, sometimes it doesn’t. For example, extension fixup had _too much_ traffic, so eventually had to be split into separate bugs on specific issues.

Hopefully, though, with Evan saying the new Blogger interface will be Mozilla-focussed, their end of the problem will get fixed soon. Perhaps unfortunately, this will also weaken any move towards fixing Mozilla’s behaviour.

(I hope this comment is fairly coherent. I wrote it in parts, heavily edited it over time and haven’t bothered to reread it. Oh, and I posted your info about </noscript> to 99647.)

 
Comment by noone #
2003-02-05 10:26:08

guys, while the software of the likes of http://www.proxomitron.org function as a localproxy, effectively filtering all the junk, I am more then happy and stupid online proxy’s don’t interest me.

once the browsers (and operating systems) start going 64bit commercially and god-knows-who will be spying on you in realtime then we’ll be in trouble.

until then bill gates’ windowsXP included spyware can kiss my @ss

btw:
”anonymous comments are allowed, so don’t bother leaving a fake email address”

what kind of sense does that make? err. no sense?

 
Comment by Phil Ringnalda #
2003-02-05 10:47:13

That means ”unlike many weblog comment systems, which require an email address, which results in you making up a fake email address because you don’t want your real address harvested, which results in me emailling that fake address if it’s not fake enough, or not emailling another address because it looks faked, my system does not require an email address, so if you don’t want to give me a real one then don’t bother leaving a fake one.” Or, as simply as possible, if you comment here again, do not enter noone@nowhere.com in the email address entry field.

 
Comment by Bernie Dodge #
2003-04-07 12:16:30

I googled my way to this discussion because I’ve been wrestling with my department’s blog which has that same thing… at least I think it’s the same.

Every time I open up the template in IE (not Mozilla) to make a change, the tag has be changed to </noscript>. If I change whatever I was going to change in the template and leave the buggered up tag that way, then nothing shows up after that point on the page when rendered. The content’s all there and can be seen in source, but some extra or missing tag shows nothing but white space after the .

So my workaround is to rechange </nospace> to every time before I save. Seems stupid.

But again, I’m confused, as this has nothing to do with Mozilla. I use IE for editing my blogs (and for almost nothing else). Can you shed any light on this?

Comment by Phil Ringnalda #
2003-04-07 12:24:30

And then to be even more confusing, the comments convert your escaped tag, and strip your unescaped tag.

So, when you edit a template it IE, the textarea contains &lt;/noscript>? What version of IE on what operating system (he asked, fearing that the answer would be some Mac version that doesn’t properly unescape entities in textareas)? It works for me in IE6/Win.

 
 
Trackback by Among Other Things #
2002-12-08 23:55:21

If this were anyone else…

…I’d soon expect to see an angry rant on the Blogger user support group. Is it a bad thing to want to make a second group, probably called ”WeanYourselfOffOfBlogger”? I mean, shit, I don’t even use Blogger that much anymore, and I feel sorry for the …

 
Trackback by phil ringnalda dot com #
2003-04-02 23:59:32

Day 122: Blogger’s still an ignorant slut

Blogger templates with noscript tags plus Mozilla equals tears

 
Trackback by phil ringnalda dot com #
2003-04-03 21:19:29

Blogger, you brilliant sweetheart

Blogger’s templates are now a Mozilla-safe zone.

 
Trackback by Jonathon Delacour #
2004-08-03 07:14:42

The possibilities of (Firefox) search

To become aware of the possiblity of the search is to be onto something. (Walker Percy) I first heard about Mozilla Phoenix/Firebird/Firefox from a September 2002 Phil Ringalda a post titled My Next Browser?: Posting from Phoenix 0.1, a re-casting of …

 
Name (required)
E-mail (required - never shown publicly)
URI
Your Comment (smaller size | larger size)
You may use <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <del datetime="" cite=""> <dd> <dl> <dt> <em> <i> <ins datetime="" cite=""> <kbd> <li> <ol> <p> <pre> <q cite=""> <samp> <strong> <sub> <sup> <ul> in your comment.