Now that site has some serious CSS. :)
Interestingly OmniWeb falls back to a simpler version.
Posted by Ben Hines at April 14, 2003 9:20 PMI suppose it would be bad for Apple to just issue a little note or even a redirect in Safari pointing people to the Guardian, instead? ;)
Posted by Scott Rogers at April 14, 2003 9:34 PMOK, I have reduced one of the rendering bugs on the iht.com website to a simple example.
http://omlc.ogi.edu/software/ihtbar.html
Only the zero is visible in Camino and IE. Hope this helps.
Posted by Scott Prahl at April 14, 2003 9:49 PMTore it up a little bit. Check out the supplied URL. You can see in the source that they actually include the same article three times, and just shift its position so that each column reveals only the next bit of text for you to read. It's absurd! I made some bits of text bold so you can see that it is the same article repeated again in each column.
Posted by Sol at April 14, 2003 9:51 PMI mean the one when you click my name. *sigh* Here it is again: http://www.avannaproductions.com/iht/
Posted by Sol at April 14, 2003 9:52 PMScott: the main bug is with the article pages. Try to reduce that, the bug is much more severe.
Posted by Ben Hines at April 14, 2003 9:57 PMWell, the nav bar item comes down to this class:
.clippingNumCol {position:absolute; height:8px; width:7px; overflow:hidden; top:9px;}
So it's the dreaded overflow bug. Just in case you were wondering about that one. Will look at the article one.
Posted by Sandy Smith at April 14, 2003 10:05 PMTo me (trying with Konqueror CVS, actually, but is prolly sufficiently close to be useful) it seems to come down to this (let's see how it will come out):
<div style="height:200px; overflow:hidden; width:738px; position:relative;">
<div style="display:block; position:absolute; width:30px;" >
text text text text text text text texttext text text texttext text text texttext text text texttext text text texttext text text texttext text text text
</div>
</div>
The JavaScript stuff seems to create 3 of those from the article text and position them... Plus there is tons of cruft to keep them positined, to move around, etc. That site really seems to hammer both Konqueror and Mozilla heavily. Disclaimer: my knowledge of CSS is close to nil, I am just reducing incrementally.
Posted by Sad Eagle at April 14, 2003 10:32 PMYes, the article bug also appears to be the lack of overflow:hidden support, if you change the line in ihtGlobal_v2.css:
#articleParent {height:200px; overflow:hidden; width:738px; margin-left:25px; z-index:1; position:relative;}
to
#articleParent {height:200px; width:738px; margin-left:25px; z-index:1; position:relative;}
Camino looks like safari.
More broken sites!
Yes, it's kind of clever. Overflow: hidden is the culprit, but their clever bit is this:
http://www.iht.com/scripts/ihtArticle.js
In it, the function setArticleHeight sets the height of the articleParent id to the height of the browser. setSnap() does the actual calculation. I'm impressed that works as well as it does across browsers. ;-)
Posted by Sandy Smith at April 14, 2003 10:43 PMAlso, there seems to be another problem with the "top" argument not working right (at least if the value is negative...) - the second repeat of the text should be 32px higher, which is exactly two lines. Works fine in IE but in Safari the offset is only slightly bigger than a single line...
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head>
<style type="text/css">
#articleParent {height:32px; overflow:hidden; width:738px; margin-left:25px; z-index:1; position:relative;}
.artCol {width:230px; position:absolute; margin-right:14px; z-index:1}
.artText {position:absolute; width:230px; top:0px; left:0px; margin-right:5px;}
.artText {font:10px Arial, san-serif;line-height:16px; color:#444444; display:none;}
</style<>
</head>
<body>
<div id="articleParent">
<div id="ac0" class="artCol"><div id="at0" class="artText" style="display: block; font-size: 12px; line-height: 16px; top: 0pt; cursor: pointer; z-index: 5; width: 230px; left: 0pt;">
Since we did change the height of articleParent to 32px which is equal to two times line-height, we should only see two lines here...
</div>
<div id="ac1" class="artCol"><div id="at1" class="artText" style="display: block; font-size: 12px; line-height: 16px; top: -32px; left: 246px; width: 230px; cursor: default;">
Since we did change the height of articleParent to 32px which is equal to two times line-height, we should only see two lines here...
</div>
</body>
</html>
Can we all just take a moment and say that, despite the fact that it has cruft and whatnot, and despite the fact that it breaks our favorite browser, whoever did that site was a mad, all-get-out genius?
I *love* clipping news articles.
Posted by nate at April 14, 2003 10:55 PMOk, I fixed a bug with overflow:hidden, and that top clipping thing is behaving now. Thanks for the reduction.
I think all that's happening with the articles is that a JS error is getting hit on some line of Javascript, and so the page is not finishing its construction (so it doesn't finish setting things up).
Someone could try narrowing that down by using a browser like Camino to save HTML complete and then manually edit the scripts, sprinkling alerts throughout. You should be able to figure out what line is confusing Safari.
AsianAvenue's site seem to misrender too, esp. the member's site
Posted by AsianAvenue at April 14, 2003 11:45 PMHi,
The IHT seem to use a similar technique to the one used by the Glasgow Evening Times. Links:
http://www.eveningtimes.co.uk/
http://www.eveningtimes.co.uk/hi/news/5014645.html
Being a big news reader, and being French at the same time, I can bring to this Comment section the following words:
Not only the IHT, but also Le Monde, our French equivalent of the Times, renders badly. I also remember problems with the nytimes.com. The Independent.co.uk displays fine, same for NPR.org by the way.
Posted by François at April 15, 2003 3:16 AMHere's a demo, info and the javascript source code behind the IHT articles:
http://www.smokinggun.com/code/sg_layout.php
I've written to the (very, very talented) guy behind the site and told him about this discussion so hopefully he'll drop by and check out what's been written so far and he might be able to offer some insight as to what's causing the mis-rendering.
My money is on this being 90% a CSS "overflow" problem.
Posted by Jeremy at April 15, 2003 3:50 AMOn a similar but different note:
A right floated paragraph:
Indent of 30 pixels: float right
ignores the text-indent setting but add an anchor into the text of the paragraph:
Indent of 30 pixels: float right with anchor
and the indent amount appears on the right.
Is this weird or is it just me?
Posted by DanB at April 15, 2003 4:55 AMSorry - here's the HTML:
A right floated paragraph:
<p style="float: right; text-indent: 30px;">
Indent of 30 pixels: float right
</p>
A right floated paragraph with anchor:
<p style="float: right; text-indent: 30px;">
Indent of 30 pixels: float right with <a>anchor</a>
</p>
Sorry for the OT, but can anyone tell me how I can get my Bookmarklets to work in Safari?
I tried to figure out how to get the selected text of the current document with JS (document.selection.createRange().text or document.getSelection()), but neither seems to work.
Also, how long is the address text in bookmarks allowed to be? Here is a comparison of other browsers:
http://www.squarefree.com/bookmarklets/browsers.html
Thanks a lot!
Fred
Not iht related but I was wondering if anybody had a work around for accessing 'Campus Pipeline' Campus Pipline is a backend that some College's use to provide online courses. Unfortunatly when you try to access it your told to take a hike and get IE 5 or Communicator. Does anybody know how to trick it into accepting Safari?
Posted by Hubert Kowalewski at April 15, 2003 7:01 AMHubert:
If it's just a browser sniffing problem, you can enable Safari's debug menu and have Safari spoof another browser. Search Google for how to enable the debug menu.
Ryan:
Thanks, I'll give it a shot.
Posted by Hubert Kowalewski at April 15, 2003 8:39 AMNot that I am complaining, as Safari progress is kicking ass!! But it would be a heck of a lot simpler to track down what was messing up a page if Safari had a better source display...
getSelection() has now been implemented (post-v73), so you will be able to use it eventually.
Posted by hyatt at April 15, 2003 11:30 AMAlex,
Bypass Safari's rather bad source display and just copy/paste it all into a new BBEdit document. Much easier to work with, and even debug.
-/-
Posted by Mikey-San at April 15, 2003 12:19 PMA little OT, but here is my interview with the designer of IHT.
http://www.ordinary-life.net/blog/archives/001117.php
It may shed some light on the problems, and it's a fun read.
Posted by James B. at April 15, 2003 2:02 PM> getSelection() has now been implemented (post-v73), so
> you will be able to use it eventually.
Thanks a lot David! I'm looking forward to beta 3 or whatever it will be.
I hope the bookmarklets can be long as well (no 508 chars restriction as in IE6)...!? Mozilla is the way to go (> 20000 chars!).
Thanks again!
Posted by Frederik Seiffert at April 15, 2003 2:10 PMSo does anyone know if it's possible to change the Google search to something else? Shouldn't this be in a config or .plist file somewhere?
Posted by pb at April 15, 2003 2:57 PMIf you feel up to the (colossal) task of configuring privoxy or some other content-manipulating/filtering proxy, you could always provide a public mirror for it. Of course, the next day, the iht.com legal team will come to take your first born. ;)
Posted by dave at April 15, 2003 3:08 PMRe bookmarklets... you know, first time I came across them when Google provided one which allowed you to select text and click the button and it would automatically search. Since then I created a bunch of others that worked great in IE6 on a Mac. When I started uusing my Mac as a primary machine, I completely had to change my workflow because they didn't work well in IE5.
Heh, and alll this time I had no clue that there was a name for these things... bookmarklets!
Thanks for mentioning them!
Posted by Nick at April 15, 2003 3:51 PMRE: Bookmarklets
If you're referring to the function within MoveableType (the same software the Hyatt uses for his 'blog), I'm having a problem with that myself. I use MT for our intranet, and Safari will not execute a javascript command directly from a bookmarklet link. That is, there is no HTML in the link, just a javascript:(YaddaYadda) command. It works fine in Mozilla, but Safari does nothing when clicking on that link.
Len
Posted by Len Laughridge at April 15, 2003 10:19 PMGreetings,
Hey ya'll, I am afraid that I am the one responsible for the IHT.
I am pretty damn busy right now, but I may be able to help out some. Just drop me a note and I will do what I can.
I have not done much JS/DOM/CSS coding in the last year or so, so I am a bit rusty..but I will try my best.
Anyway this appears to be an overflow problem, as was said before.
- jw
could one of you clever people create a bookmarklet to take the current selection and search for it at thesaurus.reference.com? of course, we'll have to wait until Safari b3 is released but it will be worth the wait.
Posted by Gregory at April 16, 2003 12:38 AMNew Safari Crashes on Launch.
Sorry, I know this isn't the place for this but Safari can't launch so I can't submit it in the browser, and I don't feel like spamming David with email so I'm posting it here.
No idea what the problem is:
**********
Date/Time: 2003-04-16 01:25:08 -0700
OS Version: 10.2.5 (Build 6L29)
Host: localhost
Command: Safari
PID: 1408
Exception: EXC_BAD_ACCESS (0x0001)
Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000000
Thread 0 Crashed:
#0 0x9068ba54 in objc_msgSend
#1 0x00acc0e0 in +[PithHelmet load] (PithHelmet.m:49)
#2 0x9068a8c0 in _objc_bindClassIfNeeded
#3 0x8fe170a8 in call_funcs_for_add_image
#4 0x8fe16fe0 in call_registered_funcs_for_add_images_in_objects
#5 0x8fe16ef8 in call_registered_funcs_for_add_images
#6 0x8fe106d8 in link_in_need_modules
#7 0x8fe121a4 in _dyld_link_module
#8 0x90016ae8 in NSLinkModule
#9 0x90689238 in objc_loadModule
#10 0x908309b8 in _NSBundleLoadCode
#11 0x90829990 in -[NSBundle load]
#12 0x00ab8bf4 in +[SIMBL loadBundle:forApplicationIdentifier:] (SIMBL.m:61)
#13 0x00ab88b4 in +[SIMBL load] (SIMBL.m:30)
#14 0x9068a8c0 in _objc_bindClassIfNeeded
#15 0x8fe170a8 in call_funcs_for_add_image
#16 0x8fe16fe0 in call_registered_funcs_for_add_images_in_objects
#17 0x8fe16ef8 in call_registered_funcs_for_add_images
#18 0x8fe106d8 in link_in_need_modules
#19 0x8fe121a4 in _dyld_link_module
#20 0x90016ae8 in NSLinkModule
#21 0x90689238 in objc_loadModule
#22 0x908309b8 in _NSBundleLoadCode
#23 0x90829990 in -[NSBundle load]
#24 0x90831c98 in -[NSBundle principalClass]
#25 0x932a80dc in -[NSInputManager _loadBundle]
#26 0x932a9298 in -[NSInputManager initWithName:host:]
#27 0x93160bf4 in +[NSInputManager initialize]
#28 0x90687210 in _destroyInitializingClassList
#29 0x906873bc in _class_lookupMethodAndLoadCache
#30 0x9068bb28 in objc_msgSend
#31 0x931602cc in -[NSApplication finishLaunching]
#32 0x931602cc in -[NSApplication finishLaunching]
#33 0x930b1be0 in -[NSApplication run]
#34 0x9315fc58 in NSApplicationMain
#35 0x000042f0 in 0x42f0
#36 0x00004170 in 0x4170
Thread 1:
#0 0x90073c48 in mach_msg_trap
#1 0x90005f90 in mach_msg
#2 0x901489f0 in __CFRunLoopRun
#3 0x90180f58 in CFRunLoopRunSpecific
#4 0x907f5a0c in -[NSRunLoop runMode:beforeDate:]
#5 0x90809244 in -[NSRunLoop run]
#6 0x7e368290 in +[WebFileDatabase _syncLoop:]
#7 0x9081cd38 in forkThreadForFunction
#8 0x90020d48 in _pthread_body
Thread 2:
#0 0x9000bc6c in getdirentries
#1 0x9000bbc4 in readdir
#2 0x9001f928 in fts_build
#3 0x9001f098 in fts_read
#4 0x7e38bea0 in WebLRUFileListRebuildFileDataUsingRootDirectory
#5 0x7e368094 in -[WebFileDatabase(WebFileDatabasePrivate) _createLRUList:]
#6 0x9081cd38 in forkThreadForFunction
#7 0x90020d48 in _pthread_body
PPC Thread State:
srr0: 0x9068ba54 srr1: 0x0200f030 vrsave: 0x00000000
xer: 0x00000000 lr: 0x00acc0e0 ctr: 0x9068ba3c mq: 0x00000000
r0: 0x00acc0e0 r1: 0xbfffd8b0 r2: 0x00000093 r3: 0x00acf158
r4: 0x9068d51c r5: 0x20000000 r6: 0xbfffd590 r7: 0x20000000
r8: 0x000ee010 r9: 0x00acf018 r10: 0x00000004 r11: 0x906faef4
r12: 0x00000000 r13: 0xa309f6b8 r14: 0xa30a08bc r15: 0xa30a08bc
r16: 0xa3098d4c r17: 0xa3098d4c r18: 0x001d5a80 r19: 0x00000000
r20: 0x00000000 r21: 0x001d0090 r22: 0x00000000 r23: 0x001bcee0
r24: 0x00acf3c0 r25: 0x00000003 r26: 0x00acf3fc r27: 0x00000001
r28: 0x00acf3f0 r29: 0x00acf128 r30: 0xbfffd8b0 r31: 0x00acc098
Razzak: http://bugreporter.apple.com/ ... learn it, use it, love it.
If you know enough to dig out a crash log you should be submitting bugs that way. Sign up for a free ADC account at http://developer.apple.com/membership/online.html and then submit the bug.
When you file bugs there, they go into Apple's system and get tracked until they're fixed. Also, Apple engineers will be able to ask you for additional information without posting in the comments in Dave's blog, which I'm sure most don't bother to read. :-)
Posted by drew at April 16, 2003 6:21 AMRazzak, shouldn't you be trying to launch Safari without the unsanctioned safari extras such as PithHelmet before logging a crash bug report? Dave has already commented that some of the distortion bugs reported by various people are probably caused by leftovers from those safari extras. Try launching Safari with the extras turned off or removed.
Posted by Gregory at April 16, 2003 7:08 AMRe: Bookmarklets
There are hundreds of sites out there providing bookmarklets for (almost) everything.
Searching at google will give you one for thesaurus.reference.com as well:
http://dictionary.reference.com/help/linking/bookmarklets.html
Posted by Frederik Seiffert at April 16, 2003 7:26 AMI was just wondering if Apple's Safari work had made it's way back into the KHTML source code yet? Anyone know? The Konqueror mailing lists don't seem to indicate.
Anyone know if the win32-khtml project or anything similar is actually getting traction? "Porting" Konq/Safari/whatever to Windows seems like a no-brainer.
Posted by pb at April 16, 2003 8:25 AMDunno if this is related to the problem, but the pages at http://www-2.cs.cmu.edu/~dst/Narconon/ are completely screwed up. They appear fine in Mozilla, but Safari doesn't show any of the pages content.
Try the above link, then choose any option on the "wheel" to see what I mean.
Posted by A random guy at April 16, 2003 11:15 AMAh, so looking at how John Weir works, that explains how the IHT site has ampersand goofs on one line and correct entites on another. I guess I cannot blame John for those scrapes across my mental chalkboard.
John Weir's "about smokinggun" page is another, possibly smaller, example of the overflow problem.
Posted by Charles Gaudette at April 16, 2003 4:08 PMquote:
Thread 0 Crashed:
#0 0x9068ba54 in objc_msgSend
#1 0x00acc0e0 in +[PithHelmet load] (PithHelmet.m:49)
#2 0x9068a8c0 in _objc_bindClassIfNeeded
Notice that "PithHelmet" in there? I'd be willing to bet this crash isn't Safari's fault.
Posted by David Smith at April 16, 2003 6:33 PMI think Apple should release a version of Safari for Linux and Windows XP. What a great way to put Apple in everyone's face. That along with iTunes. That will drive developers to the Mac platform,
Posted by jack at April 17, 2003 10:16 AMApple shouldn't do it but someone should. I'm surprised noone's jumped on this yet. A chance to build an important, widely used piece of software would seem to be attractive. There is a win32-khtml project on sourceforge that is apparently going nowhere (like 99% of sourceforge!).
Posted by pb at April 17, 2003 10:35 AMI'm also seeing overflow bugs in some sites we're developing, and it seems to be the same issue with IHT. If I have time to reduce some of our own examples, I'll post them.
If you're not thrilled with Safari's built-in View Source option, try my AppleScripts: http://webgraph.com/usability/applescripts/
They also work for all other major OS X browsers, and can place the source in BBEdit, TextEdit, PageSpinner, emacs, vi and pico.
Posted by Dan Rubin at April 17, 2003 3:17 PMBTW, there are some major display bugs we're running into over and over again when styling lists to display horizontally (inline) in Safari. Every other browser works fine (OS 9, OS X, Windows). I've submitted bug reports through Safari as of the first version, but since nothing has changed I was just wondering how I should go about drawing it to the Safari team's attention directly. I suppose I could always have Mark Pilgrim add it to his list of bugs, but it's really irking our development team since we've got a bunch of sites that use the technique, and a few more in development, and they look great in every browser but Safari.
Thanks for any suggestions -- I'll probably start posting examples (real-world and reduced) to my site soon.
Posted by Dan Rubin at April 17, 2003 3:21 PMAs others have commented, Safari's View Source option could do with being improved. A standard NSTextView is just a little bit cheap, don't you think? :-)
Posted by Kenneth MacArthur at April 17, 2003 3:53 PMI'd kinda prefer it opened in an editor, TextEdit or other.
Posted by pb at April 17, 2003 5:00 PMDan, reduced test cases and examples are what I need to fix the bugs.
Speaking of reduced test and examples - Safari behaves wrong in quite a few of the test of the HTML4 test suite...
http://www.w3.org/MarkUp/Test/HTML401/current/tests/index.html
Posted by Andreas Amann at April 18, 2003 10:39 AMWhy does Safari crash on this website?
Posted by Nick at April 18, 2003 8:13 PMNick, there's a bug somewhere in WebCore's CSS parser which causes it to crash on certain @page commands in @media print.
We're getting the same crash in OmniWeb 4.5... I might look into it someday, or I might just ask Dave to please take a look at it.
Dave, could you please figure out why linking this style sheet crashes v73?
@media print {
@page { margin: 10% }
H1, H2, H3,
H4, H5, H6 { page-break-after: avoid; page-break-inside: avoid }
BLOCKQUOTE,
UL, OL, DL { page-break-before: avoid }
}
Thanks,
-Tim Omernick
Engineer, The Omni Group
Yes, known issue. One of my top bugs right now. :)
Looking at this URL, the title should be in the font "Chiller". It is in all Mac webbrowsers, including Safari v67. But in v73, it shows in Helvetica! Why?
http://bmpmovies.kalleboo.com/index.php?show=details&movie=csteel
Posted by kalleboo at April 19, 2003 2:27 AMhttp://nslog.com/archives/2003/04/19/movabletype_searchreplace_and_fuck.php - see comments for possible Safari bug.
Posted by Erik J. Barzeski at April 19, 2003 12:55 PMIt looks like it's just about time for me to finally say CSS will be worth using, since most browsers now support it and work with most of it correctly. I guess I need to learn it.
I'm big on hand coding things, so don't care how to do it in dreamweaver or golive, etc...
Anyway, thanks tons for the work you guys are doing on Safari (Dave) and OmniWeb (Tim). I'd love to see some of OmniWeb's features in Safari's prefs, or maybe the new OmniWeb will be more like a "Safari Pro" so to speak, since it's going to be based off of WebCore.
This brings me to my other major question for Dave. Is WebCore and the Javascript equiv going to be put into the system, so that the browser can be a small shell that controls it, as can the help viewer, and OmniWeb and other apps to come, allowing the system to update it in one place and be done with it, and keeping redundancy down, or will it always have to be built into each app that uses it?
As was said earlier, via simple deconstruction (since I already mentioned that I didn't know CSS well), the overflow bit appears to be the issue.
Wow this is long winded and more off topic than I usally get...time for me to shut up. :)
Posted by Dave SPOOF Hemenway at April 19, 2003 12:57 PMThanks Tim and thanks Dave!!!
Posted by Nick at April 20, 2003 1:12 PMIn follow-up to Hubert Kowalewski's question about Campus Pipeline, even without the Debug menu there SHOULD be another page that is used to login to CP other than the site you are getting kicked out from. Open IE, enter your standard URL and see if it bumps to another page PAST the standard Pipeline URL. In our case we can simply enter "http://myscsu.southernct.edu/cp/home/loginf" to bypass any/all browser checks.
Unfortunately, we are STILL not able to utilize Safari for Campus Pipeline's Calendar and Groups functionality as it immediately logs the session out upon accessing these functions.
I'd love to be able to deploy a solid alternative to IE in to our university Mac labs and faculty, but with Camino only marginally operational and Safari unable to access 2/3 of Pipeline's areas it looks like an all M$ internet for the time being.
I've adjusted every concievable option within Safari to get it working with CP to no avail. The real kicker being that the Konqueror browser in the latest version of KDE appears to access the Calendar/Groups functions just fine. Could there be a newer version of the underlying engine in KDE/Konqueror than Safari?
Posted by Clif Hirtle at April 21, 2003 10:50 AM