log

Some lovin’ and hatin’ for the Safari 4 beta today. This is just a cur­so­ry inspec­tion from con­sumer and devel­op­er angles:

• Nice visu­als on the top sites & book­marks fea­tures. Lim­it­ed long-last­ing appeal for essen­tial­ly a glam­our fea­ture though, and it was­n’t imme­di­ate­ly appar­ent how to edit the top sites, where they were stored, and how they came to get in my brows­er (I had the same data cus­tomi­sa­tion issues with Flock a while back). Oper­a’s Speed­Di­al feels much more user-friend­ly at this stage.
• Tab place­ment is hor­ren­dous on Win­dows; at least Chrome’s tabs make a pass at being in the brows­er win­dow’s con­tent area. Stop screw­ing with inter­face con­ven­tions — treat­ing some tab mouse events as the app title bar is con­tex­tu­al­ly con­fus­ing (dou­ble-click, drag­ging). Please stop inno­vat­ing things that don’t need innovation.
• Each Safari release reminds me how good Webkit’s ren­der­er is. Awe­some.
• Thanks for leav­ing the abil­i­ty to chose alter­nate font ren­der­ing in — I thought this might be on the chop­ping block. I hate XP’s ren­der­ing of gener­ic web type­faces. Switch­ing between the default and alter­nate actu­al­ly exposed the cause of the head­er posi­tion­ing bug on my Test Design 26 page, heh.
• Please imple­ment mid­dle-click tab close.

I was sur­prised to see CSS Ani­ma­tions and CSS Effects men­tioned on the Fea­tures page — these are Apple CSS specs that (as far as I know) are still exposed in Webkit with the ‑webkit- pre­fix. The CSS Work­ing Group site lists these mod­ules as an upcom­ing work­ing draft (unlinked but avail­able on the w3c dev serv­er). Are they mature enough for use out­side of an closed, inter­nal ecosys­tem? I must admit that I empathise with any­one who tries to get specs through the W3C process.

Speak­ing of specs in progress, I was delight­ed to see the Safari site’s writ­ten in HTML5 (yes I’m a hyp­ocrit for advo­cat­ing unfin­ished specs here and ques­tion­ing them above, but HTML5 is much more mature). IE demands sac­ri­fice, so the page includes

<script type="text/javascript" charset="utf-8">
	document.createElement("article");
	document.createElement("nav");
	document.createElement("section");
	document.createElement("header");
	document.createElement("aside");
	document.createElement("footer");
</script>

to make non-exis­tent tags be treat­ed as both non-void and styleable. The Safari start­up page actu­al­ly uses the <video> and <audio> tags too, and it sure was freaky to see video with­out Flash installed.

[Update: eek, the page isn’t valid HTML5. <sec­tion>, <head­er>, <foot­er> and <nav> are mis­used. For the <foot­er> links they’d be bet­ter off using a par­ent UL instead of <h3>s. Not sure what’s trig­ger­ing the <script> error though.]

Great quote from Ian Hick­son dur­ing the dis­cus­sion on Rob Sayre’s reduced-scope spec:

> You’re the dic­ta­tor. Am I read­ing the WHATWG process wrong?

Appar­ent­ly. A dic­ta­tor, by def­i­n­i­tion, has total pow­er. I have near­ly no pow­er; I am con­trained by lega­cy con­tent, by the whims of imple­men­ta­tors, by ratio­nal and log­i­cal argu­ment, by the needs of authors and users, and by research.

On anoth­er note, Anne van Kesteren links to an awe­some CSS thread which pro­vides much oppor­tu­ni­ty for facepalm­ing to any prag­ma­tists reading.

(Clean­ing up an unfin­ished-post queue, for those won­der­ing. This one’s a lit­tle skimpy but sta­tis­tics are always fun.)

A while back I held a small work­shop at TAFE on brows­er usage and the impor­tance of ver­i­fy­ing designs in mul­ti­ple browsers. Lucky I don’t get sick of com­plain­ing about IE6; even now, sev­en years after its release, mar­ket share keeps it on our test­ing matrix:

• IE6: ~36%
• IE7: ~32%
• Fire­fox: ~19%
• Safari: ~3%
• Opera: ~2%

(Stats tal­lied from sev­er­al pub­lic sources late last year)

Of course these stats are a rough aver­age from sev­er­al sources wide­ly vary­ing demo­graph­ics (and the num­bers don’t even add up to 100%).

One of the few seri­ous ben­e­fits of MS’s Vista push is IE6 being errad­i­cat­ed in its wake. XP’s lat­est ser­vice pack still does­n’t include IE7 though — they must be fac­ing off with a lot of cor­po­rate intranet admins whose busi­ness soft­ware still depends on IE6 Javascript bugs.

Curi­ous­ly, Microsoft will be ‘push­ing’ IE8 to Vista and XP via OEM chan­nels (think Dell) as an option­al com­po­nent. Maybe web devs should start peti­tion­ing OEMs to enable it by default? Imag­ine writ­ing only one stylesheet per website.

Using the delight­ful­ly inflam­ma­to­ry title “Stu­dents to be taught there’s no God”:

http://www.news.com.au/story/0,27574,24797395–29277,00.html

I remain respect­ful­ly cyn­i­cal of any organ­i­sa­tion preach­ing eth­i­cal stand­points to chil­dren in the absense of parental super­vi­sion, but I might be a lit­tle less cyn­i­cal in this case; like many cen­tre-left thinkers (read: aca­d­e­m­ic snob) I believe break­ing the cycle of the all-see­ing author­i­tar­i­an father-fig­ure is a wor­thy goal. The ener­gy we inject into feed­ing reli­gion could be bet­ter spent on edu­cat­ing the incom­ing gen­er­a­tion on the need for com­mon moral­i­ty and respect built from our own expe­ri­ences, not from sources irra­tional­ly divi­sive and sci­en­tif­i­cal­ly and soci­o­log­i­cal­ly thousounds of years out-of-date.

To be blunt I’ve always seen San­ta Claus as a micro­cosm for the greater reli­gious direc­tion: as irra­tional, inex­pe­ri­enced chil­dren with no under­stand­ing of the mores of larg­er soci­ety, most of us need a struc­ture to keep us in check until we learn to ratio­nalise on our own and out­grow the need to per­form for the omnipresent being. Have we as adults ever out­grown the next phase?

(This bypass­es the entire dis­cus­sion on whether young chil­dren are even sen­tient, which is inter­est­ing in its own right.)

On the prob­lem of export­ing ideals and com­mon­al­i­ty to chil­dren, I think we’ve lost a lot by becom­ing a media-dri­ven cul­ture with­out the face-to-face fam­i­ly and com­mu­ni­ty dis­course that ties peo­ple togeth­er through shared expe­ri­ences. And I under­stand that said dis­course is a core part of the reli­gious expe­ri­ence; I’d like that same dis­course but with­out the par­rot­ing of ancient verse and the attri­bu­tion of our moral struc­ture to mys­ti­cal beings. It gen­uine­ly speaks of our will­ing­ness to demote our poten­tial. Do we actu­al­ly want to improve? That involves grow­ing up as a soci­ety and drop­ping a lot of bag­gage, and sad­ly for many peo­ple that bag­gage pro­vides the com­mon­al­i­ty that allows us to speak to one anoth­er civil­ly, since most oth­er avenues of bond­ing are closed off.

As a race we’re intrin­si­cal­ly attract­ed to sto­ry­telling and the want to believe in some­thing greater than our­selves; I think that’s because we nom­i­nate any height we can’t yet reach for idol­i­sa­tion. Maybe that ten­dan­cy is inescapable with­out a change of wiring… but a dis­cus­sion of vol­un­tary human­i­tar­i­an eugen­ics and genet­ic engi­neer­ing is some­thing I’ll leave for anoth­er post.

A long stand­ing prob­lem with graph­ic files on the web is that there’s no lossy for­mat with trans­paren­cy options. Three options are incom­ing, all real­is­ti­cal­ly long-term if you’re devel­op­ing for the entire web and not one par­tic­u­lar platform.

JPEG + SVG Fil­ters in HTML = masked lossy images. Allows for both raster and vec­tor mask­ing, but only for Fire­fox 3.1 for now. This brings the addi­tion­al ben­e­fits of the entire SVG fil­ter line­up avail­able any­where in HTML.

Webkit’s CSS Masks applic­a­ble to any visu­al ele­ment entire­ly using CSS pro­ter­ties. Allows for a raster mask or ref­er­enc­ing an SVG. Isn’t as exten­sive as the SVG Bling approach — it’s intend­ed pure­ly for mask­ing — but has low­er bar­ri­er for entry.

JPEG-XR, a new JPEG for­mat seem­ing­ly unre­lat­ed to the orig­i­nal bar the name. High­er qual­i­ty, small­er file­size, options for loss­less encod­ing, and an alpha chan­nel. Con­spir­acists may ques­tion Microsoft­’s involve­ment after the OpenXML trav­es­ty but as long as licenc­ing is free and adop­tion high­er than the wavelet-encod­ed JPEG stan­dard from a few years back, I’m happy.

That which rates a last men­tion is Microsoft­’s Com­pos­i­tor fil­ter. The doc­u­men­ta­tion hints at being able to per­form cal­cu­la­tions using an alpha chan­nel, but I’ve no idea if that trans­lates to final ren­dered trans­paren­cy since the exam­ples are too com­plex for me to spend time decyphering.

A while back I cre­at­ed a test design called Sans & Serif that need­ed a large high-colour trans­par­ent image with alpha blend­ing (the leaves). PNG 32 did the job, but it’s non-lossy so the file size was ~210k. I end­ed up con­vert­ing the leaves to 8‑bit with some visu­al trick­ery to hide the aliased edges. I’m look­ing for­ward to doing this properly.

T.V.Raman:

there’s a poten­tial cri­sis for the W3C if we say “HTML 5 is so dif­fi­cult to deal with that we’ll ignore it” then much of the work in W3C becomes irrel­e­vant to deploy­ment on the Web