· Technology
 · · Web
 · · · Services

Learning REST · I was talk­ing to a col­league who has to be­come savvy in a hur­ry about mod­ern Web ap­pli­ca­tions and he asked “How do I learn about REST?” Good ques­tion. I thought of a cou­ple of sug­ges­tion­s, then asked Twit­ter and got some more. Here they are ...
The Shambling WS-Undead · I’ll try to play this straight. It seems that a posse of in­dus­try ti­tans (IBM, Or­a­cle, CA, and EMC) want a W3C work­ing group to stan­dard­ize WS-Transfer, WS-ResourceTransfer, WS-Enumeration and WS-MetadataExchange. Be­cause, as they say, “There is still some work to be done”, and “Accessing da­ta about a re­source through Web ser­vices is an area of the Web ser­vices ar­chi­tec­ture that has yet to be ful­ly realized.” I guess that if you re­al­ly do want to im­ple­ment HTTP on top of the SOAP stack on top of HTTP, these are clear­ly the Right Ven­dors For The Job. There is, how­ev­er, re­al dan­ger in this move, as out­lined by Mark Not­ting­ham in The WS-Empire Strikes Back­... fee­bly.
RESTful JavaOne · I on­ly man­aged to take in a few talks here and there, but the ones I did catch sure had some first-rate REST preach­ing. To the ex­tent that there’s a sur­pris­ing trend at this year’s J1, I’d say it’s REST-is-in, “Big Web Services” (see be­low) are out ...
WS-Enterprise, 2008 · I was talk­ing to a very large financial-sector com­pa­ny in the mid­dle of Amer­i­ca in the mid­dle of last week, and the cog­ni­tive dis­so­nance around this whole WS-thing is pal­pa­ble ...
On Web Service Definition · There’s a flur­ry of dis­cus­sion about whether or not, in the world of REST, you need some sort of for­mal spec­i­fi­ca­tion for the ser­vices you’re of­fer­ing. My con­clu­sion: yes, but in a very application-specific way ...
WS-dämmerung · Snell: Now that I’m work­ing for IBM’s We­bA­head group, build­ing and sup­port­ing ap­pli­ca­tions that are be­ing used by tens of thou­sands of my fel­low IBMer­s, I haven’t come across a sin­gle use case where WS-* would be a suit­able fit. Obasan­jo: The on­ly times I en­counter some­one with good things to say about WS-* is if it is their job to pimp these tech­nolo­gies or they have al­ready “invested” in WS-* and want to de­fend that in­vest­men­t. Vi­nos­ki: Fi­nal­ly, I re­al­ized that WS-* was sim­ply not worth it to any cus­tomer or to me. I re­mem­ber the days when it was ba­si­cal­ly just Mark Bak­er and me shout­ing “The WS-King has no WS-clothes and there are WS-bleeding-sores on his WS-butt!” The easy ad­vice for the CIOs and CTOs of the world is “Just don’t buy that crap”. The more dif­fi­cult ad­vice is “In fu­ture, care­ful­ly con­sid­er the mo­tives and world-views of those who were try­ing to con­vince you to buy that crap.” It’s OK, CIOs and CTOs don’t read on­go­ing. Any­how, we’re still a mi­nor­i­ty; I get email ev­ery day from peo­ple ad­ver­tis­ing prod­ucts and ser­vices and cours­es and ad­vice on “SOA Governance”. Mo­tives and world-views, re­mem­ber?
Tab Sweep — Tech · Au­gust is sup­posed to be the slow time of year. Not! Is there ev­er a lot of in­ter­est­ing stuff out there. To­day we have WS-funnies, OOXML Pur­dah, Web names, In­ter­net Registry struc­tures, and Ru­by metapro­gram­ming crazi­ness ...
Tech Tab Sweep · We’re all over the map to­day, from gen­er­al the­o­ries of soft­ware de­vel­op­ment to low-level op­ti­mized bit-banging. Wel­l, all over the soft­ware map, I guess ...
REST News · I’d been go­ing to do a REST news round-up, but Dave John­son beat me to it. He calls it “Atom and REST” and yes, I’ve no­ticed a pat­tern where you get a hand­ful or more of RESTa­far­i­ans in a room, vir­tu­al or re­al, and Atom keeps bub­bling up. Not just the pro­to­col, but the da­ta for­mat too. Oop­s, Dave missed one: Pete Lacey ...
WCF-REST · It’s like this: The WS-* project’s at­tempt to re-invent RPC and pre­tend that you can suc­cess­ful­ly take an object-model view of net­worked ap­pli­ca­tions looks in­creas­ing­ly fan­ci­ful, in the gen­er­al case. On the oth­er hand, Microsoft’s Mer­ry Men slaved away on Indigo, worked around the hor­rors of XSD and WSDL, built some pret­ty good Visu­al Stu­dio tool­ing, and shipped DCOM, the Next Gen­er­a­tion WCF; now it’s the way Win­dows wants to be talked to over the net. Which is why the Ja­va ecosys­tem has things like WSIT, built in­to JAX-WS; you may not like Win­dows but ev­ery­one has to talk to it ...
The London Illustrated News · I spent the week in Lon­don. Fun was had, pic­tures were tak­en, I learned things. Here­with il­lus­trat­ed notes on trans­porta­tion, en­er­gy, fi­nance tech­nol­o­gy, busi­nesslike drink­ing, women’s cloth­ing, Groovy, ex­cel­lent lamb-chop cur­ry, and a round red anoma­ly ...
WS-* in the Springtime, O Joy · This hap­pens over and over. New WS-* spec sub­mis­sion, check. In­sane­ly huge char­ter lock­ing down the con­clu­sion and en­sur­ing a rubber-stamp out­come, check. Loads of de­pen­den­cies on WS-standards, WS-drafts, WS-submissions, and oth­er WS-handwaving, check. Res­o­lute obliv­i­ous­ness to oth­er tech­nolo­gies that ad­dress the same prob­lem, check. [Up­date: Hos­til­i­ties break out in the com­ments. Read ’em and see what you think.] ...
Tech Tab Sweep · I break with my no-underlying-theme theme and do an all-technology tab sweep; in fac­t, al­most all XML ...
Udell and Vinoski · I al­most nev­er lis­ten to pod­cast­s. I don’t com­mute, and just sit­ting in front of my com­put­er, I tend to get dis­tract­ed. But I did man­age to hang in through all but the trail­ing cou­ple of min­utes of Jon Udell’s A con­ver­sa­tion with Steve Vi­nos­ki about ser­vices, the en­ter­prise, and the web. Since I’m called out by name, I think I should prob­a­bly re­spond ...
SOA, REST, Java · What I usu­al­ly do in this space is grum­ble and whine about SOA & SOAP and so on. To­day, let’s start with a laugh, in­stead. But ac­tu­al­ly, there’s news; the Ja­va tribe has de­cid­ed to take REST se­ri­ous­ly, see JSR 311 and, for some more point­er­s, Eduardo’s write-up. I haven’t had time to do a deep-dive, but I’m re­as­sured by the pres­ence of Marc Hadley A.K.A. the WADL guy; and it looks that there will be oth­er re­al ex­perts at the table. On the oth­er hand, El­liotte Rusty Harold emits an ex­tend­ed snarl. (Sam­ple: “I hope we can de­rail this completely...”) Hey El­liot­te, I guess mak­ing friends and in­flu­enc­ing peo­ple is for loser­s, right? The proof of the pud­ding, ob­vi­ous­ly, is in the eat­ing, but the fact that this dis­cus­sion is hap­pen­ing has to be a good thing ...
SOA and WCF · More dark clouds gath­er; storm sig­nals in­clude the gen­er­al trash­ing of the whole WS-* stack by Gartner’s Nick Gall, the con­tin­u­ing broad­sides (lat­est here) from Pete Lacey, and Give It a REST, a sol­id piece of ar­gu­ment from Lar­ry O’Brien. But I think the re­al take-away, while a lit­tle sub­tler than “WS-* is broken”, is be­com­ing pret­ty ob­vi­ous ...
WS-Tilkov · Ste­fan Tilkov is emerg­ing as one of our best thinkers on the whole WS-Sanity fron­t. I rec­om­mend a (slightly-old) In­foQ video in­ter­view, a man­i­festo on his own site called 10 Prin­ci­ples of SOA, and (a­gain at In­foQ) an in­ter­view with Pete (“S for Simple”) Lacey. Hm­m, seems like I’m point­ing to In­foQ a lot late­ly.
[1 comment]  
S for Simple · I feel guilty some­times about the lull in my WS-Rants, be­cause the forces of WS-Complexity and WS-Darkness are out there evan­ge­liz­ing tire­less­ly. But to­day I feel bet­ter, be­cause there are pow­er­ful WS-dialogues out there speak­ing truth to con­fu­sion. Dun­can Cragg has pub­lished Get­ting Da­ta and Set­ting Da­ta, which he says are the first two of a nine-part (!) se­ries en­ti­tled The REST Dia­logues. Quite sound and in­sight­ful, I think. But laugh­ter is di­vine and the di­vine trumps the ra­tio­nal; in that spir­it I rec­om­mend Pete Lacey’s The S stands for Sim­ple, which is in a class by it­self. [Up­date: DHH piles on.] [Up­date: Nel­son Mi­nar too.] [Up­date: and Sam Ru­by]. [Up­date: Lacey’s follow-up, They can’t hear you, is a must-read.]
Erroneous Ministerial One · Here­with my oc­ca­sion­al romp through the built-up brows­er tab­s. Item (se­ri­ous): In The ‘Next’ Ja­va, Joe Gre­go­rio says some Real­ly Smart Things about lan­guages in gen­er­al and Ja­va in ar­tic­u­lar. Item (se­ri­ous): At Busi­ness Week, Stephen Baker’s Writ­ing for an au­di­ence of one says some­thing gen­uine­ly new (hard, these days) about blog­ging. Item (in­ter­est­ing): My new Sam­sung is a pret­ty cool phone, but there are a few ir­ri­tants. It turns out that some­one called RedIpS has fixed them. I just bought a flash­ing ca­ble on EBay; I won­der if I’m go­ing to be break­ing any laws? Item (not se­ri­ous): SOA Facts. Item (puz­zling): Some guy named Tim Bray seems to be in trou­ble in Chi­na; this ar­ti­cle pro­vid­ed the ti­tle above. I hope Mr. Bray gets out OK.
[1 comment]  
How We Learn · Here’s the bald truth: the state of the art in In­for­ma­tion Tech­nol­o­gy is be­ing ad­vanced, first, in re­leas­es of open-source tech­nol­o­gy (which speak loud­er than word­s) and, in­so­far as words go, pri­mar­i­ly in on­line site-to-site con­ver­sa­tion­s. You can watch it hap­pen. Michi Hen­ning hauled our body of knowl­edge one small but im­por­tant step up the end­less moun­tain­side with his The Rise and Fall of CORBA, in the always-excellent ACM Queue. Bruce Eck­el, in Are Web Ser­vices Real? Part II, fo­cus­es on the ob­vi­ous pro­cess par­al­lels be­tween CORBA and WS-* (re­li­able laugh line: “WS-* is be­com­ing CORBA, on­ly with an­gle brack­ets to make it slower”). Fi­nal­ly, Steve Loughran’s On Cor­ba, DCOM, ICE, and dis­tribut­ed ob­jects in gen­er­al re­al­ly goes deep, won­der­ing whether dis­tribut­ed ob­jects are an in­her­ent­ly bro­ken idea. His clos­ing word­s: “REST han­dles it best by freez­ing the set of verbs to a low num­ber, on­ly al­low­ing one way links, but at a price, the price of no easy map­ping be­tween REST re­sources and na­tive class­es, no two-way links and (cur­rent­ly) not very easy APIs. The ques­tion is, when will the En­ter­priseys no­tice that this is the on­ly thing that has been shown to work.” I don’t think the “Enterprisey” ep­i­thet has been a very use­ful ad­di­tion to our dis­course; but aside from that, well, yeah.
Bold Strides · I’m glad to see that the Web 2.0 com­mu­ni­ty is com­ing to un­der­stand the ben­e­fits of Service-Oriented Ar­chi­tec­tures. I re­fer to SOA in­te­gra­tion with Flickr and del.i­cio.us, which demon­strates an out­stand­ing­ly ma­ture ap­proach.
The End of SOA · I did an in­ter­view and a pod­cast [Up­date: here] at that Rails con­fer­ence and the ques­tion came up in both, and in the hall­way talk too: “What do you think we should do about SOA?” Which weird­ly, no­body had asked me be­fore, and I could find on­ly one an­swer: “Don’t do any­thing. ‘SOA’ may have meant some­thing once but it’s just ven­dor bull­shit now.” Look­ing back, what hap­pened was, cer­tain soft­ware ar­chi­tects were un­com­fort­able with the fram­ing that goes with the words “Web Services”; maybe be­cause peo­ple think any­thing with “Web” in the name should be sim­ple and lightweight and easy to set up. Thus SOA, which is so much more En­ter­prisey. Me, I want to go the oth­er way. The cru­cial point is that Web-like things should be sim­ple and lightweight and easy to set up; so I think the “Web” part of “Web Services” is more im­por­tant than the “Services” part. SOA isn’t the fu­ture, Web style is.
Important, I Think · I’m gen­uine­ly para­noid about bang­ing my own drum and shout­ing “Listen to me!” be­cause I know how of­ten I’ve been wrong about things, and how much of the fu­ture is de­ter­mined by luck and raw ran­dom chance. That said, if the lessons I’ve learned over the years mean any­thing, there’s a con­ver­sa­tion go­ing on right now that’s re­al im­por­tan­t. Here are three start­ing points: Go­ing Down To The Cross­roads from Don Box, Styles: Beyond WS and REST from me, and Spend­ing the $100 from Don. They aren’t the whole con­ver­sa­tion, but they re­flect it well and have point­ers to most of the rest. Right now, a lot of peo­ple think that Web-flavored frame­works are the fu­ture of heterogeneous-network ap­pli­ca­tion­s, which is to say al­most all ap­pli­ca­tion­s; and that the WS-mountain is re­al­ly a WS-molehill; and that we need to fix up the tool­ing for de­vel­op­er­s. Depend­ing how much pull Don has, Mi­crosoft might be first off the mark; fair enough. But I re­al­ly think this de­serves at­ten­tion. In an in­ter­est­ing side­light, Rob Sayre (in com­ments here) and Dare Obasan­jo have agreed with Sam Ru­by that if you’re build­ing an ac­tu­al ap­pli­ca­tion us­ing Web-flavor APIs, well, by gol­ly, you ought to play by the Web-Architecture rules. Glad you guys think so.
Just Contractors · I think David Heine­meier Hansson’s graph­ic is worth a link; the ti­tle kind of puz­zled me, but the com­ments have a point­er to an ex­pla­na­tion.
Too Much Fun · That’s what Dave Or­chard is hav­ing; check out WS-SopranosDesperateHousewivesKwisatchHaderach. Al­so out there in the “that much fun must be illegal” ter­ri­to­ry is David Isen­berg, he’s build­ing a whole fran­chise around “Fat Pipe. Al­ways On. Get Out of the Way!” cre­do; there are now bumper stick­ers and T-shirts. He sent me some; thanks, David. (But see David Weinberger’s wor­ry).
Styles: Beyond WS and REST · There’s been a re­cent mini-flurry around REST and al­leged sub­sets such as “Lo-REST” and “Hi-REST”. Ini­tia­tor: Don Box. Re­spons­es: Obasan­jo, Jon­nay, Glazkov, Meg­gin­son, and To­mayko. My thanks to all of them for keep­ing this stuff in the front of my mind. I’m not sure that “Web Services” and “REST” are use­ful names for the in­ter­est­ing network-application styles. But I’m pret­ty sure I know what those styles are ...
WS-Fritz-Lang · You know, I think most peo­ple would agree that I’ve writ­ten quite enough about this stuff. But this one par­tic­u­lar WS-piñata swung right in­to my face, and I sim­ply can’t re­sist a whack at it. In this par­tic­u­lar case, I even have some of the po­lit­i­cal back-story. But first, check out Evolv­ing Web ser­vices stan­dards for man­ag­ing sys­tem re­sources and have a peek at this re­mark­able PDF white pa­per: Toward Con­verg­ing Web Ser­vice Stan­dards for Re­sources, Events, and Man­age­ment (both re­leased March 15 by HP, IBM, In­tel, and Mi­crosoft­) ...
WS-Crossroads · Check out Web Ser­vices at a Cross­roads, by Daryl Plum­mer, who’s group VP and chief fel­low at Gart­ner. Yet an­oth­er find­ing that when it comes to Web Ser­vices, you can choose be­tween sim­ple,re­li­able, standards-based in­fras­truc­ture that’s here now and the sprawl­ing, shift­ing WS-* tech­nol­o­gy that’s still un­der con­struc­tion, most­ly by IBM and Mi­crosoft, both well-known cham­pi­ons of sim­plic­i­ty. Plummer’s piece is most­ly right, but I’m go­ing to ig­nore Tolstoy’s ad­vice and pick some nit­s ...
WS-Angst · There’s been an in­ter­est­ing flur­ry of high-level WS-discussion, launched by Don Box in Prag­mat­ics (it’s short, go read it, the long string of com­ments doesn’t add much). The dis­cus­sion sloshed around the bl­o­go­sphere; I’ll pick some high­light­s. Ste­fan Tilkov says Mu: “I do be­lieve that on a very high lev­el, the de­bate is ut­ter­ly irrelevant.” I don’t. I just don’t be­lieve that there’s a lev­el high enough that large-scale ba­sic in­fras­truc­ture bets don’t mat­ter. Chris Fer­ris of­fers con­ven­tion­al wis­dom: WS-* is just fine, tak­ing a lit­tle longer than we’d hope­d, but you’re re­al­ly gonna need this stuff. Pa­trick Lo­gan push­es back: “I was shocked how lit­tle in­ter­op­er­abil­i­ty, not to men­tion func­tion­al­i­ty, has been ac­com­plished in the WSDL and SOAP world over the last sev­er­al years ... we de­cid­ed to run a dif­fer­ent ex­per­i­men­t... ad­dress the same busi­ness prob­lem but with just HTTP. Very soon we were spend­ing all our time talk­ing about busi­ness func­tion­al­i­ty and mes­sages rather than in­fras­truc­ture headaches.” Oo­h. Dare Obasan­jo weighs in twice: More on Prag­ma­tism and Web Ser­vices, and es­pe­cial­ly Why WS-* in­terop sucks. I’ll leave the last word to Rob Sayre in No, It’s Over, I Real­ly Mean It: “If you have Mi­crosoft say­ing ‘well, the best ap­proach is to make this elab­o­rate in­fras­truc­ture we’ve spent bil­lions of dol­lars build­ing out optional’, then the de­bate is over.” Me, I think the WS-stench of some­thing WS-rotting from the WS-head down is be­com­ing in­creas­ing­ly dif­fi­cult to ig­nore.
WSbert · It’s been a cou­ple of months since I’ve hit the WS-punching-bag, but one can hard­ly ig­nore it when WS-jargon creeps in­to the cyn­i­cal geek’s holy of holies; that’s right, I’m talk­ing about Dil­bert. In oth­er re­cent news, last week SDFo­rum put on an In­ter­op­er­abil­i­ty Work­shop, nice­ly writ­ten up by Paul Krill and Eve Maler. There were some ter­rif­ic sound-bites: “I un­der­stand what all this stuff is and it still makes my head spin”; “The ven­dors are al­ways pur­su­ing their own agenda” (I’m shocked, shocked); “You re­al­ly do need to be a rock­et sci­en­tist to use a lot of it”; and “The thing we have to be a lit­tle care­ful about is that we need ab­strac­tions that don’t as­sume that you can cov­er up com­plex­i­ty with tooling”. But you know what, maybe Dil­bert re­al­ly is more in­dica­tive; when you have start­up com­pa­ny, you know you’re in trou­ble when Dil­bert car­toons start show­ing up on cu­bi­cle wall­s. Let’s see, in this cor­ner we have Gart­ner say­ing that if you’re not do­ing WS-strategic-synergies (as in WS-big-budget), you’re In The Wrong Quad­ran­t. In the oth­er cor­ner, we have Dil­bert. I know who I’d bet on.
The Aquarium · Peo­ple have been ask­ing for it since forever, and re­cent re­leas­es of Roller now have a group-blogging fea­ture. I’ve al­ways been skep­ti­cal of the idea, which re­de­fines “blog” from be­ing about an in­di­vid­u­al voice to be­ing about a shared in­ter­est; be­cause peo­ple are big­ger and more in­ter­est­ing than their in­ter­est­s. But over the last cou­ple of week­s, my at­ten­tion has sev­er­al times been drawn to pieces in The Aquar­i­um, which is a group blog about GlassFish. While I’m not re­al­ly an EE kin­da guy, I can see how some­one who cares about that stuff could find this kind of a re­source use­ful. For ex­am­ple, check out this point­er to an over-elaborated but nonethe­less use­ful ar­ti­cle about do­ing REST in in JAX. Maybe “group blogs” have legs.
WS-Interop · We did a Web Ser­vices an­nounce­ment Fri­day, and while I’m not on that team, I think it’s sig­nif­i­can­t. Ba­si­cal­ly, it says that we’re im­ple­ment­ing enough of WS-* to in­ter­op­er­ate with Mi­crosoft Indi­go Win­dows Com­mu­ni­ca­tions Foun­da­tion, and that the im­ple­men­ta­tions will be Open Source. Let’s be blunt (and re­mem­ber, I speak on­ly for my­self here): WS-* isn’t about stan­dard­s. It’s about what Mi­crosoft (there are part­ner­s, but it’s most­ly Mi­crosoft) choos­es to im­ple­ment while wav­ing the WS-banner and retroac­tive­ly shak­ing Stan­dards Fairy Dust over the pro­cess. Which is OK, as far as it goes; I get the im­pres­sion that Indi­go WCF is ac­tu­al­ly some pret­ty neat soft­ware that will be use­ful to a lot of Mi­crosoft cus­tomer­s, and Sun has a stake in the ground say­ing we’re go­ing to in­ter­op­er­ate with the Mi­crosoft WS-stack. Do I think this stuff is go­ing to Change The World? No. Do I think that this is the re­al fu­ture of Web Ser­vices? No. (I think the fu­ture looks like Ama­zon Web Ser­vices or the Atom Pub­lish­ing Pro­to­col or even re­cent ar­rivals like the Back­pack API, even though Back­pack could be done with the Atom Pro­to­col). But I do think that WS-Microsoft-Interop is the right thing to do at the mo­men­t. Oh, and by the way; if you’re one of the many many peo­ple grind­ing away at one of the WS-inventions that hasn’t shown up al­ready in this in­terop sto­ry, well, there may be grounds for wor­ry.
WS-Stardate 2005.10 · “How’s the WS-* field strength, Mr. Spock?”
“Steady at 783; sub-optimal, but manageable.”
“My in­tu­ition tells me something’s wrong.”
“All right, I’ll run a deep scan, but...”
“Captain! I’m get­ting a weird read­ing from three specs in the Se­cu­ri­ty sec­tor; it looks like...” [A weird shaft of bril­liant pur­ple light stabs through the bridge, fry­ing the red-shirted en­sign where he sit­s.]
“Mr. Spock! What was that?”
“Checking, Cap­tain; those WS-warbirds are Se­cureCon­ver­sa­tion, Trust, and Se­cu­ri­tyPol­i­cy. They’ve been there for years, but some­how they’re dif­fer­en­t... aaaah. They’re de­ploy­ing an OASIS-TC stan­dard­iza­tion field!”
“But that’s a friend­ly tac­tic, Spock.”
“No, Cap­tain, they’re mod­u­lat­ing the field with a locked-down char­ter de­vice; the TC has to just ap­prove them the way they are.”
“That’s fiendish!”
“And ex­treme­ly il­log­i­cal, Captain.”
[Sud­den­ly the bridge rocks and the lights flick­er.] “Engineering! Scot­ty! What’s happening?”
“Cap’n, I din­na un­der­stand it, they’re growing!”
“How can that be... Spock?”
“He’s right, Cap­tain, they’re us­ing the superseded-spec maneuver.”
“Scotty, do we have the bandwidth?”
“I dun­no Cap’n, SecureConversation’s been su­per­seded from 17 to 31 pages, and Trust from 41 to 68.”
“My God, they’re grow­ing like can­cer. Scot­ty, I need more bandwidth!”
“We’re doin’ our best, Cap’n... Aaaaaaaaagh!
“Report, Scotty!”
“Cap’n, cap’n, it’s Se­cu­ri­tyPol­i­cy, curse it... lurk­ing at 13 pages since 2002 with­out a peep, but there’s a su­per­sede; it’s up to 90. Cap’n... she can­na take any more. She’s gonna blow!

Check out Jon  · Jon Udell is an ex­is­tence proof of the need for tech­nol­o­gy writ­ers who are tech­ni­cal­ly com­pe­tent but don’t have a non-writing day job; in an ide­al world this is how all tech jour­nal­ists would be. First this great big honkin’ sur­vey piece on Web Ser­vices; I’ve been feel­ing guilty about not cov­er­ing that ter­ri­to­ry more, but now I don’t have to be­cause Jon is. Sum­ma­ry: There is hope. Then, his ex­cel­lent in­ter­view with Bill Gates, in which Gates is in­for­mal, in­for­ma­tive and in­tel­li­gen­t, as op­posed to Ballmer’s party-line blovi­a­tion.
WS-Reality at RouteOne · Last week at Ja­va One, Ashesh Badani, a Sun SOA mar­ket­ing per­son, want­ed to have lunch with me to talk about WS-*. He brought along T.N. Subra­ma­ni­am, Direc­tor of Tech­nol­o­gy for RouteOne, a car-loan ag­gre­ga­tor. (Sun loves RouteOne, they’re a ref­er­ence cus­tomer not on­ly for us but for SeeBeyond, which we’re in the pro­cess of ac­quir­ing). Any­how, nei­ther Ashesh nor Ashok Mollin, a Sun guy who’s been en­gaged at RouteOne, got a chance to say much, be­cause T.N. and I hit it off and had a good time talk­ing about Web Ser­vices. Which RouteOne are do­ing, big time and for big bucks and suc­cess­ful­ly. They are ex­act­ly the kind of peo­ple that those of us strug­gling in the WS-* morass ought to be look­ing to for lesson­s. This, I think, will be the first ev­er on­go­ing piece struc­tured as an in­ter­view; with T.N.’s help, I’ve tried to re­con­struct our con­ver­sa­tion at lunch. I think some con­clu­sions are ob­vi­ous, but I’ll leave them for you to draw ...
WSD-Survey · Norm and I post­ing our NSDL and SMEX-D pro­pos­als seems to have un­leashed a flood of en­er­gy in this space. I pre­vi­ous­ly point­ed to Dion Hinchcliffe’s sur­vey work; well, Dion is re­al­ly get­ting down to busi­ness with his Tak­ing Stock of Web Ser­vices De­scrip­tion. He’s go­ing to be tak­ing a se­ri­ous run at four­teen (!) dif­fer­ent can­di­date lan­guages, ap­ply­ing them to a re­al web ser­vice and do­ing re­al im­ple­men­ta­tion­s. I don’t see any­thing to com­plain about in his ap­proach. I’m sub­scribed, you betcha, and if you care about Web Ser­vices you should be too. For­tu­nate­ly, Dion is braver than I and sup­ports com­ments, so there may be some in­ter­est­ing di­a­logue there.
Raining on the Parade · I guess it’s good that Steve and Scott made nice, and there’s no doubt that when the cus­tomers tell you to in­ter­op­er­ate, then you bloody well in­ter­op­er­ate, so it was a good piece of work (see Pat Patterson’s take in a com­ment on his own blog). But this glue for link­ing to Microsoft’s WS-Federation is a second-rate so­lu­tion at best. Among oth­er rea­son­s, WS-Federation is yet an­oth­er WS-backroom spec that might change (or go away) any time the peo­ple in the back­room want it to; not some­thing I’d ad­vise bet­ting on. If you have prod­ucts from any two ven­dors that im­ple­ment Lib­er­ty Al­liance specs prop­er­ly, well, they in­ter­op­er­ate. Sin­gle sign-on? Yawn. Pret­ty well ev­ery­body is a mem­ber, oh ex­cept Mi­crosoft. If the cus­tomers want sin­gle sign-on (and they do want sin­gle sign-on), Mi­crosoft should bloody well join Lib­er­ty and im­ple­ment the spec­s, then they’ll have in­ter­op­er­a­tion with ev­ery­one, not just Sun.
The Web Service Description Space · I’ve been mean­ing to go back and up­date my ear­li­er piece on re­plac­ing WSDL, but Dion Hinch­cliffe has pub­lished a roundup of all known Web Ser­vice De­scrip­tion al­ter­na­tives, with a pic­ture even. Dion is clear­ly now the world’s lead­ing au­thor­i­ty on this field.
Replacing WSDL, Twice · Let’s make three as­sump­tion­s: First, that Web Ser­vices are im­por­tan­t. Se­cond, that to make Web Ser­vices use­ful, you need some sort of dec­la­ra­tion mech­a­nis­m. Third, that WSDL and WSDL 2, de­spite be­ing the work of re­al­ly smart peo­ple, are so com­plex and ab­stract that they have un­ac­cept­ably poor ease-of-use. What then? Nat­u­ral­ly, the mind turns to a small­er, sim­pler suc­ces­sor, sac­ri­fic­ing gen­er­al­i­ty and es­chew­ing ab­strac­tion; in ex­act­ly the same way that XML was a suc­ces­sor for SGML. Well any­how, that’s the di­rec­tion my mind turned. So did Norm Walsh’s; his pro­pos­al for NSDL al­so in­cludes a help­ful ex­pla­na­tion of why Web-Service de­scrip­tion is im­por­tan­t. My sketch is called SMEX-D. In­ter­est­ing­ly, NSDL and SMEX-D, al­though both wave the ban­ner of The Sim­plest Thing That Could Pos­si­bly Work, are wild­ly dif­fer­en­t; NSDL is the sim­plest way you could pos­si­bly de­clare an RPC-style func­tion call with po­si­tion­al pa­ram­e­ter­s. SMEX-D is the sim­plest pos­si­ble way you could de­clare an ex­change of XML mes­sages. Which is more im­por­tan­t? Are both nec­es­sary? I sus­pect that these days, the Sim­plest Thing That Could Pos­si­bly Work would in­clude a dec­la­ra­tion that a par­tic­u­lar message-exchange/function-call should be re­li­able, us­ing HTTPLR or equiv­a­len­t. Are there any oth­er pro­pos­als or skunkworks float­ing around out there? Let me know and I’ll ag­gre­gate point­er­s. [Up­dat­ed with more pro­pos­al­s, startling com­men­tary from Meg­gin­son and Obasan­jo, and an ap­peal to Sowa’s law]. ...
SMEX-D · SMEX stands for Sim­ple Mes­sage Ex­change, and SMEX-D for SMEX De­scrip­tor, an XML lan­guage de­signed to pro­vide sim­ple de­scrip­tions of a wide range of Web-Service mes­sage ex­changes, both REST-based and SOAP-based ...
WS in BW · A month or two ago, there was a piece about Web Ser­vices by Jim Ker­stet­ter in Busi­ness Week, and it was pret­ty good but I dis­agreed with one or two of his premis­es, so I wrote to tell him so. He an­swered, say­ing “Want to write a view­point piece on the sub­ject for us?” So I did, it’s on­line here.
SOA Top or Bottom? · John Crupi writes that SOA de­ploy­ments should be top-down, “problem to ar­chi­tec­ture to solution”. He specif­i­cal­ly says that wrap­ping ex­ist­ing tech­nol­o­gy de­ploy­ments in a Web Ser­vices wrap­per is a “perfect recipe for a SOA failure”. Hm­m... these are strong claim­s, rad­i­cal I think, and fur­ther­more, quite new to me. And thus worth point­ing to.
Web Services: Spring 2005 Roundup · The usu­al Web Ser­vices back­ground rum­ble has been get­ting re­mark­ably loud these past few week­s. [Why now? -Ed. Beats me. -T] I haven’t fig­ured out where we’re go­ing, but then no­body else has ei­ther. This frag­ment sur­veys the state of play with a tra­di­tion­al link-roundup, and con­cludes with a sug­ges­tion that those who care vis­it Ja­pan in May. [Up­dat­ed with a point­er to an an­a­lyt­i­cal piece from Joshua Al­len.] ...
Stop WSDM · Over at OASIS, they’re work­ing on YAWSS (Yet Another Web-Services Spec) called WSDM. The com­mit­tee de­cid­ed they were done and asked for an OASIS-wide vote; the re­sult was 67 yes, 7 no. In­ter­est­ing­ly, the 7 “No” votes weren’t about the sub­stance of WSDM, they were about the fact that it has de­pen­den­cies on all sorts of oth­er WS-bric-a-brac that isn’t fi­nal­ized yet, in­clud­ing a W3C Sub­mis­sion and a bunch of oth­er com­mit­tee draft­s. The com­mit­tee pon­dered this and de­cid­ed to go ahead and make it a stan­dard any­how. I tried to go and read WSDM and it made my head hurt, severe­ly; it’s gnarly and huge and com­pli­cat­ed and seems to de­pend on lots of oth­er gnarly and huge and com­pli­cat­ed things. So, any­one who wants to im­ple­ment this is go­ing to have to make a ma­jor in­vest­men­t, and since a lot of the rel­e­vant specs are un­sta­ble, you just know some part of that in­vest­ment is go­ing to get thrown on the trash-heap. In­ter­op­er­abil­i­ty? Ha. Ha. Ha. This suck­s. I don’t want to be an ab­so­lutist here; some or­ga­ni­za­tion­s, like IETF, to­tal­ly for­bid this kind of thing while oth­er­s, like ISO, al­low them in a kind of con­trolled way. But in this par­tic­u­lar case, what they’re try­ing to do is deeply wrong and the OASIS man­age­ment needs to find a way to stomp on it if they want to re­tain any cred­i­bil­i­ty. For oth­er com­men­tary, start here. [Dis­clo­sure: I don’t un­der­stand WSDM and I don’t even un­der­stand the prob­lem it’s try­ing to solve and while Sun was one of the “No” vot­er­s, it was strict­ly on the de­pen­den­cies is­sue, and I don’t know whether we, cor­po­rate­ly, are as ir­ri­tat­ed as I am in­di­vid­u­al­ly and I don’t know whether we, cor­po­rate­ly, ac­tu­al­ly care about this tech­nol­o­gy and if we do, whether we like it or not.]
Another Loyal Oppositionist · I haven’t been bang­ing my WS-Gadfly drum re­cent­ly, but that’s OK, be­cause James Gover­nor is on the job.
Real World Web Services · This is an O’Reilly book by Will Iver­son, whom I don’t know. Giv­en my fre­quent pub­lic grum­bling on the sub­jec­t, I thought I should give it a se­ri­ous look ...
UBL by the Numbers · Via Jon Bosak, a point­er to this XML 2004 pre­sen­ta­tion (Pow­erPoin­t, sigh), about the Dan­ish Government’s de­ploy­ment of a bunch of XML tech­nolo­gies in­clud­ing UBL. Check out slides 4 & 5: they es­ti­mate the an­nu­al sav­ings achiev­able from in­voic­ing in UBL at some­where be­tween €100M and €160M. I may be out of step with the crowd but it seems painful­ly ob­vi­ous to me that UBL is go­ing to be huge and I don’t un­der­stand why more tech­nol­o­gy ven­dors (in­clud­ing my em­ploy­er) aren’t re­fo­cus­ing their e-business strat­e­gy around it.
Evolution in Action · Adam Bos­worth doesn’t write very of­ten. But when he does, you re­al­ly want to read it.
Three Questions on XSD and WSDL · Last week at the Colorado Soft­ware Sum­mit, dur­ing my keynote I asked three ques­tions of the at­ten­dees, who were a few hun­dred most­ly se­nior de­vel­op­er­s, most­ly from the Ja­va ecosys­tem. (I’ve tucked a pic­ture in the body of this piece.) Do you use XML Schema? Pret­ty well ev­ery hand went up. Do you think you un­der­stand XML Schema? One hand went up. Do you like XML Schema? A scat­ter­ing of hand­s, maybe 20%. I asked the same three ques­tions about WSDL; sim­i­lar pat­tern, not quite as uni­ver­sal ex­po­sure, a few more thought they un­der­stood it. Just re­port­ing ...
WS-Pagecount · Here­with a brief re­port from the op­po­si­tion bench­es in the WS-Parliament. My re­cent piece in­tro­duc­ing the “loyal opposition” idea pro­voked quite a bit of feed­back, some of which is worth high­light­ing. Al­so, those of us in the skep­tics camp have been heard to mut­ter dark­ly about “thousands of pages of specifications” and I won­dered whether those barbs were jus­ti­fied, so I had my com­put­er count ’em. Read on for the an­swer. [Up­date: I may have mis­count­ed.] [A­gain: push­back on the JSR anal­o­gy.][A­gain: Hey, they added some more!] ...
The Loyal WS-Opposition · In places that have Par­lia­men­tary sys­tem­s, the ma­jor­i­ty par­ty (or par­ties) form the Govern­men­t, and the rest form the Op­po­si­tion. In con­sti­tu­tion­al monar­chies the jar­gon is “Her Majesty’s Loy­al Opposition.” The idea is, they Op­pose the Govern­ment but are Loy­al in that they promise not to lead a mob with pitch­forks to string them up; and they stand ready to pro­vide an al­ter­na­tive. When it comes to Web Ser­vices, that’s what I’d like to be: the loy­al op­po­si­tion ...
What Adam Said · A month or two ago, I got a call out of the blue from some guy I nev­er heard at a re­cruit­ing shop say­ing “I’m look­ing for a ref­er­ence on Adam Bosworth.” I’m afraid I gave the guy a hard time, first I took him for a prankster and then I rant­ed at him along the lines of “You’re talk­ing about prob­a­bly one of top twen­ty soft­ware peo­ple in the world, have you nev­er heard of Qu­at­tro Pro and Mi­crosoft Ac­cess and IE 4? What are you talk­ing to me for?” To his cred­it, the guy was pa­tient and ex­plained that this was Google and Google is dif­fer­en­t. So I told him about my ex­po­sure to Adam over the years and all the things it’s painful­ly ob­vi­ous that he’s good at, and I’m sure the oth­er peo­ple they called did too, and now he’s at Google. But what I re­al­ly want­ed to say to­day is, go read his lat­est piece, he says in one para­graph what I’ve been rav­ing about for months in the area of Web Ser­vices: “The re­al­ly use­ful things turn out to be the sim­plest ones.” How many times do we have to re-learn this lesson? Any­how, there’s more, and it’s all good.
WS-Sanity that Fits In Your Hand · Check out Nokia Web Ser­vices Frame­work for De­vices — a Service-oriented Ar­chi­tec­ture. It’s a prac­ti­cal in­tro to how SOA might play in the mo­bile space, with some em­i­nent­ly sen­si­ble back­ground work; there’s a sec­tion en­ti­tled What is a service-oriented ar­chi­tec­ture, and why is it good?. Any­one who doesn’t think that the cen­ter of grav­i­ty of net­worked com­put­ing, which means all com­put­ing, is mov­ing to­ward the mo­bile space and isn’t read­ing Rus­sell Beat­tie and prob­a­bly should be. As I’ve said be­fore, if you want to know the fu­ture of SOAs and Web Ser­vices, look at the peo­ple who are ac­tu­al­ly de­ploy­ing them and see what they’re do­ing. Any­thing they’re not go­ing to use, you can ig­nore for now.
From the Web · I was talk­ing to Mark Hap­n­er, a smart guy here at Sun who does heavy Ja­va ar­chi­tec­ture, about WS-Sanity, and he had an an­gle that’s new to me. In re­cent decades, he points out, good new tech­nolo­gies have first ap­peared in rough-and-ready form on the In­ter­net, then mi­grat­ed in­to the en­ter­prise. (I re­mem­ber when query track­ing first showed up on the FedEx site; that was ten years ago, and it in­stant­ly opened a few mil­lion eyes to a bet­ter new way to de­liv­er data). But all the WS-* hul­la­baloo is try­ing to go the oth­er way; it’s try­ing to mod­el all the (nec­es­sary) com­plex­i­ties of cur­rent IT in­fras­truc­ture and turn them in­to many thick lay­ers of ab­strac­tions wrapped around a Web­bish core. So, if you be­lieved in his­to­ry, where would you look for the fu­ture of “Web Services?” You’d look at the peo­ple who are do­ing them in a rough-and-ready fash­ion out there on the Net. The names that come to my mind are Ama­zon, Google, EBay, Sales­force.­com, maybe SABRE. [Up­date: Via the Sabre Geek, a point­er to what they’re do­ing.] What­ev­er they’re do­ing, that’s Web Ser­vices or SOA or the Ser­vices Fabric or what­ev­er you want to call it. Any­thing they don’t need, maybe it isn’t go­ing to be re­al im­por­tan­t.
SOA Talk · I’m lis­ten­ing to Steve Gill­mor, Doc Searl­s, Jon Udel­l, Dana Gard­ner, and Dan Far­ber talk about SOA via “The Gill­mor Gang” at ITCon­ver­sa­tions. Here­with some ob­ser­va­tions on the form and con­tent ...
Gunfight at the WS Corral · As a stu­dent of WS-Geography and WS-Politics (and there’s a lot to learn), my eye­brows were first raised by a point­er to a warmish protest (I quote: “... the draft Oa­sis Web Ser­vices Reli­a­bil­i­ty spec­i­fi­ca­tion was sav­aged in a most un­for­tu­nate man­ner... We are al­so in­formed that the IBM as­sas­si­na­tion at­tempt will be post­ed on the Oa­sis web site which fur­ther adds in­sult to injury...”) con­cern­ing a pre­sen­ta­tion en­ti­tled Crit­i­cal Com­par­i­son of WS-RM and WS-R, which is in­deed post­ed on the Oa­sis site. How­ev­er, to keep things fair, the site al­so has a point­er to the re­sponse from the WSRM TC. I to­tal­ly have no opin­ion as to who’s right, or if the prob­lem is a fruit­ful area for stan­dard­iza­tion work. But I won­der why, if there are dif­fer­ing ideas on how to solve this prob­lem, and there is a stan­dards or­ga­ni­za­tion at work, the dif­fer­ing ideas aren’t be­ing hashed out in the stan­dards or­ga­ni­za­tion. Clues may be found in the email thread be­gin­ning here. It’s tough for strangers to learn a new land­scape when it’s rav­aged by war­ring tribes.
UBL 1.0 · On May 1st, UBL 1.0 hit the street­s; to­day, Jon Bosak is in Hong Kong launch­ing the mar­ket­ing push. UBL is a set of general-purpose XML-encoded busi­ness doc­u­ments: or­der­s, ac­knowl­edg­ments, pack­ing slip­s, in­voic­es, re­ceipt­s. I’m not a UBL ex­pert, but I have two good ar­gu­ments that say it’s like­ly to be im­por­tant and suc­cess­ful ...
WS-Sanity in Ireland · Peo­ple have asked why, since I joined Sun, I’ve been go­ing on about WS-this and WS-that. In part it’s just be­cause I still care about all things XML. But most­ly, it’s be­cause in late 2003, be­fore I came here, I had an eye-opening ex­pe­ri­ence that changed how I think about Service-Oriented Ar­chi­tec­tures and Web Ser­vices. I think that the fu­ture is in plain sight, and that’s be­cause it’s be­ing built right now by the Govern­ment of Ire­land, and it’s called reachser­vices ...
Sean++ · Sean McGrath speaks wis­dom about Trans­ac­tions and SOA. Any­one who’s try­ing to find their way through the WS-Confusion would do well to pay reg­u­lar at­ten­tion to Sean. Among oth­er things, he does large SOA de­ploy­ments for a liv­ing; we’re talk­ing prac­tice not the­o­ry.
Web Services Theory and Practice · I look at the con­fus­ing land­scape where the “Web Services” flags fly, and I see some things that are proven good prac­tice, and some oth­er things that are most­ly the­o­ry. I think we could all be san­er about all this if were care­ful to dis­tin­guish be­tween the­o­ry and prac­tice ...
WS-* · Ex­tract­ed from the valu­able list at CBDI, and omit­ting those marked su­per­seded, we find BPEL4WS (Busi­ness Pro­cess Ex­e­cu­tion Lan­guage for Web Ser­vices), WS-Addressing, WS-AtomicTransaction, WS-BPEL (Web Ser­vices Busi­ness Pro­cess Ex­e­cu­tion Lan­guage), WS-CAF (Web Ser­vices Co­or­di­na­tion Frame­work, in­clud­ing WS-CTX, WS-CF, and WS-TXM), WS-Choreography, WS-Coordination, WSDL (Web Ser­vice De­scrip­tion Lan­guage), WSDM (Web Ser­vices Distribut­ed Man­age­men­t), WS-Eventing, WS-Federation (Web Ser­vices Fed­er­a­tion Lan­guage), WSIL (WS In­spec­tion Lan­guage), WS-Manageability, WS-Notification, WS-Policy, WS-Provisioning, WS Reli­able Mes­sag­ing, WS-ReliableMessaging (not the same as the pre­vi­ous), WS-RF (WS-Resource Frame­work), WSRP (WS Re­mote Por­tal­s), WS-Security, WS-SecureConversation, WS-SecurityPolicy (an ad­den­dum to WS-Security), WS Se­cu­ri­ty Ser­vices, and WS-Trust. Hey, they missed WS-MetadataExchange, and now I see that we have WS-BaseFaults and WS-ServiceGroup as of yes­ter­day. Is this the fu­ture? Is the em­per­or dressed?
Listen To Don · That Don Box, he’s a smart guy. I’m glad to hear I’m not the on­ly one wor­ry­ing that the profoundly-good Web-Services idea is dan­ger of be­ing dis­cred­it­ed by too much the­o­ry and not enough prac­tice.
author · Dad · software · colophon · rights
Random image, linked to its containing fragment

By .

I am an employee
of Amazon.com, but
the opinions expressed here
are my own, and no other party
necessarily agrees with them.

A full disclosure of my
professional interests is
on the author page.