The Old New Thing

Well, part of the reason for a "1" is so that you don't accidentally make a toll call. You can only make a toll call if you dial "1", and thus you can know the difference between making a toll and non-toll call.

So, the recording should probably be something like "The number you dialed is a toll call. You can connect now by pressing '1'. To avoid this message in the future, dial '1' before this number."

That is, inform them that it's different from what they typed, but let the call connect anyway if they desire.

I often search Google for things on Yahoo. Google is my default search engine and its convenient. Yahoo's main page doesn't have easy links to the services I use most there, like groups, travel, and stock quotes. They're still trying to be a search engine, but most people use them as a set of services.

As for the whole phone thing, I agree. It's annoying to have to dial a number 2-3 different ways just to get the right prefix. It's even more annoying when it's different between your home phone and cell phone.

Oh no, no, no, no....you can most certainly make a toll call without dialing a "1".  Several years ago I got an extra $90 on my phone bill because my dial-up ISP was a toll number on my phone plan.

"(Fortunately, the rule on when you have to dial the area code is easier to remember: If the area code you are calling is the same as the area code you are dialing from, then you can omit the area code.)"

This is no longer true in New York City, where you must always dial the full 10 digit phone number (including the area code) even if you are in the same area code.

Once King County grows as big as NYC you can share our problems!

In addition to that it is all about convenience.

It much easier to just type "yahoo" in the search box and click the link in the search results page than type in additional "www." and ".com" and then press enter.

My personal preference is to use the address bar for everything, which is why I use Maxthon. If the text in the address bar looks like a url then it would take me to the site, else it would take you to a search page you configure.

"This is no longer true in New York City, where you must always dial the full 10 digit phone number (including the area code) even if you are in the same area code."

In eastern Massachusetts, we go one better (har har), since we have to dial 1 + the 10 digit number for all calls.

The problem is the rules seem somewhat arbitrary as to when you need the one, when you don't, and when you do or don't need an area code.  It gets especially weird in places that have overlay area codes.

Of course, one answer to Raymond's "why don't they do this?" question is that the phone companies are not customer-service driven.

Oh how true.  Similarly, the "missing ';' before identifier..." error in C++.  I'm hoping some future iteration of Visual Studio (i.e. C#, C++, J#, etc.) has a compile error event that a third-party can intercept and add value (in this case, just aborting the compile/build, adding the semicolon and starting the compile again--if no other errors were found)...

For the phone number problem, I think the correct way would be to require the full number (including the country code) no matter what.

For searches, my personal preference is to enter URLs into the address bar as is, and search for things by prefixing them with "gg" for Google or "я" for Яndex.

Ultimately, the difference between an URL and a search term is that URLs are unique keys while search terms are not. In a sense, all those people search for *a yahoo*, while those other people connect to *the Yahoo*. It is just a happy coincidence for the former that the most known yahoo in the Internet is The Yahoo.

I completely agree with your UI sentiments, Raymond.  Windows should implement one of the least talked about features of Macs... Spotlight.  Sure, there's similar functionality in using indexing service, but that's always felt like a hack as it runs in userland, not at the kernel level.

Spotlight is an index of all the user generated content on the computer which is updated (I think) via the processor when any file actions occur.  Yes, this slows down processor calls, but when processor cycles are so cheap, this is a good use of them.

I made my wife get a PowerMac when she go her laptop just so I could see how this would work with a normal user.  She's horrible about organizing her file system (tending to dump everything in Documents or on her Desktop without any of the folder organization we use).  Every time she can't find a file, Spotlight rescues her with a quick (sub second) search of every file on her computer.  Definitely the most overlooked user-friendly feature I've seen in years that, as you say, conforms the computer to how people want to work, not vice-versa.

Sometimes it's pure laziness. I often use google to search for a webpage that I know the URL of. When I launch IE or Firefox Google is the default, the search bar is the control that has focus and sometimes I forget that.

When i type in the url and look up I see it in the search bar rather than the address bar. To go to the URL the 'proper' way means highlight, Ctrl+C, Click on url bar, ctrl+v, enter or I just just hit enter and click on the link.

I'm lazy :-)

D'oh my bad.  I use my cell phone too often (where the +1 is never needed).  In NYC we have to do all 11 digits as well...  Sorry to mislead people.

"Similarly, the "missing ';' before identifier..." error in C++.  I'm hoping some future iteration of Visual Studio (i.e. C#, C++, J#, etc.) has a compile error event that a third-party can intercept and add value (in this case, just aborting the compile/build, adding the semicolon and starting the compile again--if no other errors were found)..."

Oh, please no. This sort of DWIM is bad, bad, bad, especially in a programming language. 90% of the time adding the semicolon might be the right thing, but the other 10% of the time it won't be what I wanted (maybe I forgot a typedef or have a typo in a macro), and in those cases it will take me hours to find the place where the compiler hook inserted a semicolon that I didn't want.

Making life easy for users is good. Programmers should have to think.

The rules on 7/10/11 digit dialing are dictated by the state PUCs; don't blame the telcos for the stupidity of government bureaucrats.

My cell phone, which the state _doesn't_ regulate in this way, is a lot easier to use.  I can dial +1number, 1number, or just number (where number is 10 digits, due to having 6 area codes here).  The telco figures out just fine what I meant, and they'd do the same on land lines if the PUCs let them.

The whole "toll alerting" thing is a throwback to when toll calls cost several dollars per minute; requiring the 1 made sense back then.  Now, with LD charges being flat or pennies a minute, it's time for it to go, but the PUCs haven't caught up.

re: semicolon

The VAX/VMS C compiler used to do the best of both worlds: it would pretend that there was a semicolon so you could find any other errors in the result, but would mark it as an error so tht no object file was created.  This nicely reduced the compile-edit-test cycle but didn't let people get away with skipping the semicolons.

I'll go even one step further than JS Bangs: any compiler that lets missing semicolons slide without them being an error will just result in a metric hogshead of bad code that emits thousands of warning messages when compiled.

A better approach might be a hook in syntax coloring so that the problem can be fixed before the compiler gets it.

There are problems with the combined search/URL bar approach. IE 4-6 used a "search if we cant reach the URL or if it's not a URL" scheme, but it had some rather annoying issues:

* It would replace what you typed in the box with something like "http://search.msn.com/search?blablaetc." meaning that if you made a typo when entering a URL, you couldn't just go back and fix it.

* Error conditions other than "No DNS record" triggered the search, including "DNS unavailable". This meant that it would attempt to search when not connected to the internet, combined with the above issue, if you used a bookmark or link when not connected, you would end up with an error (can't find MSN search) and a broken URL, so once you connected, it still wouldn't work properly.

Also, the search engine was frequently changed by malware. With no GUI to the setting and no way to recall the default setting, you were stuck with it, even after the offending malware was removed.

> ... Maxthon ...

Um, Firefox does this by default, if I understand what you're doing.  Type something into the address bar.  If FF thinks it's a URL, it'll go there.  Otherwise, it'll send the string off to your default search engine, extract the first result, and go there.

(Or maybe it's hardcoded to send it off to Google and do an "I'm feeling lucky" search.  I'm not sure because my default search engine (for FF's search bar) is Google.)

If you mean it actually takes you to the search results page, then that's what FF's search bar does.  But its search bar doesn't recognize URLs, so maybe that's still a reason to use whatever it was that you use.

FF also automatically adds the www and com (and a bunch of other TLDs) if you omit them in the address bar.  I'm quite sure that IE (at least 6) does the same thing.

"A better approach might be a hook in syntax coloring so that the problem can be fixed before the compiler gets it."

This is what Eclipse does, and it saves lots of time correcting simple typos and other innocent syntax errors. But the Eclipse compiler is so fast that it hardly matters. I'm sure that this could be worked into VS easily.

All those who are in favor of DWIM should read this article:

http://dictionary.reference.com/browse/DWIM

There are some subtle, if minor, dangers in searching for a URL rather than entering it directly:

1) Google knows where you're going:

2) Anyone who can convince Google that they're really Yahoo can thereby capture your traffic.

In other words, Google becomes a dependency.

I think the #1 reason people search for things like yahoo.com at google is that search boxes tend to steal focus!  They probably think they're typing in the browser's address bar!

I can sympathize on the 1- dialing problem; we have the same mess here in eastern Pennsylvania.  You always have to dial at least 9 digits now, but some numbers require a 1- prefix, and some don't, and you need a table in the phone book to know which ones do and which ones don't.  My next-door neighbor?  No 1- prefix.  My father's office 45 minutes away in a different area code?  No 1- prefix.  The Best Buy store only 15 minutes away?  1- prefix.  Joy rupture.

There's a reason why I use my cell phone for a lot of calls these days; the land line rules are beyond overly complicated, growing into the land of nonsensical.

Perhaps the phone dialling is a legacy thing that can't be changed without impact existing apps. Then you would blame the phone when it is the legacy apps using the wrong +1 API's ...

Tongue in cheek here! But perhaps true. Lest not judge, unless ye yerself want to be judged ... or whatever it is ...

Stu: I actually hit that one earlier today, configuring my new NAS unit - enter 192.168.11.150 in the browser - and when the NAS doesn't respond in time, IE goes off trying to search the web for 'relevant' pages instead. Irritating.

"This is no longer true in New York City, where you must always dial the full 10 digit phone number (including the area code) even if you are in the same area code.

Once King County grows as big as NYC you can share our problems!"

This was proposed in Washington (and implemented in Oregon), but IIRC the state PUC demanded the ability to keep 7 digit dialing after the public outcry over the changes. Here at the University of Washington, though, 10 digits is still required, despite being in Seattle.

Strangely enough, I've also noticed people in UW's CS department also Google for URLs, like "washington.edu". They know the difference between search engine queries and URLs, and they can deal with errors, but for some reason, they ALWAYS go to Google. I don't understand it.

The one thing that's always confused me about the phone sytem is that what is considered "long distance" seems pretty arbitrary. For example, when I was at a conference near the Canadian border, I was able to call home toll-free by using a VoIP number registered in Whatcom county (a couple hours north of Seattle). According to DSLReports, the CO of that number is actually IN Seattle, yet it costs nothing to dial it, and some outrageous price per minute to dial a regular 206 number.

What James said. The problem with 'doing the right thing' for the user in the case of typing in the URL in the wrong box is that it penalizes the user in a lot of other valid usage scenarios.

If I type in the wrong thing, it takes at least twice as long for IE to tell me about it because it has to go find me some stuff I don't care about, rather than simply telling me what I'm looking for doesn't exist. Drives me crazy, and makes me glad I use a Mac most of the time.

Making this tradeoff in the wrong direction seems to be somewhat ingrained in MS culture, because you see the same thing with the various agents (the late Clippy and the non-late autocorrect) in Office.

Sometimes simpler is better, even if it means the user occasionally sees an error notification.

re: Tarig

>>It much easier to just type "yahoo" in the search box and click the link in the search results page than type in additional "www." and ".com" and then press enter.

Ever try ctrl-enter in the url box?  Try it -- just type yahoo then hit ctrl-enter.  Works in IE and Firefox.

That delay is really annoying.  One of the many settings I change on a new account/machine is to set "do not search from the address bar".  I don't use the IE search bar, but do have the google search bar, and it doesn't steal focus on me.

The rules for 1+ in Eastern Mass (4 regions, 2 area codes per region) are really screwy too.  Shrewsbury to Northborough (neighboring towns) is 10 digits, Northborough to Marlborough (neighboring towns) is 10 digits.  Shrewsbury to Marlborough is 1+10 digits.  It does seem to map very well to toll calls, until you start including regional plans with allowances for "nearby" toll calls.

In Western Mass (1 area code), it's now 7 digit or 1+10 digit depending on local vs. long distance.  The Verizon cell phones out there are also confused.  If you include the 413 area code, the call fails.  As far as I can tell, the phone does not know its area code, unless they've moved it to some hidden setting that only Verizon can program.

This can also be a problem in testing...I became infamous in certain quarters for avoiding the mouse.*  I use alt-d without thinking. I routinely tab between links. I use ctrl-tab and shift-F10, even.

One advantage to automated testing is that it's rote. (Yes, this is also a disadvantage.) But if you have eight different ways to do the SAME THING, and it's immensely critical that all eight different ways work, then it makes much more sense to automate all eight ways of doing it than to rely on humans to switch between all eight ways of doing it. Assuming your test does a good job of identifying failures, of course. :)

*I found wearing my doctor-prescribed wrist braces to bug triage ended a lot of the fights over whether keyboard accessibility was "really" important.  

I don't have a a problem with RC's contention that there ought to be a single box to type into and it should adapt to the user, because most users aren't geeks. But as a geek, I despise and detest the fact that I cannot type a hostname and expect on any given browser to be taken to the web server named. When instead I get a page of search results or the first hit from the search results, I get angry. I also strongly dislike the additional dependency it implies. One solution would be a separate program for geeks like me with a simple text box which is highly configurable but defaults to doing what I say rather than what it thinks I must mean. But of course that program won't be installed everywhere.

There is another possible reason that long distance calls require a 1: area codes and local exchange codes (the first three digits of a phone number) can overlap, and unlike cell phones (and URLs), landline phones have no unambiguous signal that a phone number is complete.

Also, I have not been able to use the 1+7 phone number type in many years.

Vorn

Thanks for linking to my original item about why Yahoo is the most-searched term on Google. There was a followup that pretty much agreed that navigation is answer:

http://www.networkworld.com/community/?q=node/5934

PingBack from http://geoffreyzheng.wordpress.com/2006/11/27/accessible-technology/

I hate to be offtopic...but can someone point me to a page that describes the 7/8/10/11 digit dialing rules?

One frustrating thing is that the abstraction sometimes falls through.

When typing in the address bar of IE, if it can't locate the URL I entered, it searches.  For example, I have entered www.psmk2.com, when the actual URL is www.psmk2.net.  Windows Live Search helpfully says "hey, I think you meant www.psmk2.net!"... but if I click on that link, it searches on www.psmk2.net instead of taking me there.

When I first saw it, I was amazed that it had actually done what I, as a user, would have expected it to do.  Then I clicked on the link, and got exactly what I didn't want.  It was like a bait and switch.

Note on the adding 1 message:

This is sometimes a good thing. (and the .net designers had/borrowed a good idea when they created [Obsolete("reason", true)] and [Obsolete("reason", false)])

It can actually be a major hassle maintaining backwards compatibility on something* if you want to reuse the conceptual or actual space freed up by deprecating an older mode. Far better (where doing it is better than the alternatives) to do this nicely:

1)minor version change adds obsolete(false).

2)major version change with that in place.

3)major version change with it (true).

4)major version change with it removed entirely

5)major version change using same space for new reason.

Obviously the 3rd phase is a pain if you didn't fix it at 1. going from 1 to 4 is even less pleasant and 1 to 5 is going to hurt badly. But at least in this way you reduce the unexpected failure case likelihood and pain (and any coder going up 3 major versions and not looking at the "breaking Changes" list is mad)

I get why you don't do this in OS api's (and why Raymond has high utility value to us all in doing what he does) but in other areas the trade off sometimes needs to be made, the only way to get back that space is to turf people off it, especially where this can be done internally.

Unfortunately the

"I'll break compatibility if you are compiling against me since a human should get to see the reason and take action and correct the flaw. But if you run against me at runtime I'll happily service the request"

model doesn't dovetail with the compile and run  differences.

interesting assumption.... but... .the reason i use the search bar instead of the addy bar even when i know the url is..... i hate erasing the places i commonly go which are easily found in my address bar by hiting the down arrow and a little scroll.  if i type in some insignificant url that i may have a hair up my ass to look at, it just ADDS MORE AND MORE URL's to my addy history. and eventually i have to ERASE MY HISTORY... and type in all the common places i go every day again.  using the search bar saves me time and energy. cause i get the same result, ie,. getting to the site im looking for without chumming up my shit.

Amen!  I especially hate the whole thing about how you should NOT dial a "1" before some number - why can't they just ignore it?  

Here in Dallas, you have to dial the area code even if you are calling within the same area.  I live in the 214 area so I always have to dial 214.  But guess what?  Some parts of the 214 area code are actually long distance so you have to dial a "1".  Arrghhhhh!!!!

"If I type in the wrong thing, it takes at least twice as long for IE to tell me about it because it has to go find me some stuff I don't care about, rather than simply telling me what I'm looking for doesn't exist. Drives me crazy, and makes me glad I use a Mac most of the time."

Tools > Internet Options > Advanced > Search from the Address bar > Do not search from the Address bar > OK.

Problem solved.

"Making this tradeoff in the wrong direction seems to be somewhat ingrained in MS culture, because you see the same thing with the various agents (the late Clippy and the non-late autocorrect) in Office."

In any Office program: Tools > Autocorrect Options > uncheck what you don't want.

Problem solved.

Do you even have these options to enable on a Mac if you wanted them? The "MS culture" if there is one, is to try to give 99% of users what they meant, not what they asked for. Then  provide a way to turn that feature off for the 1% of people that don't want it.

I agree with kokorozashi, when Microsoft "dumbs down" the UI they convert more geeks to Linux. I can understand why you would want to make it easier for general users, and why for many "More options confuse and limit rather than liberate", but that doesn't mean you should punish those of us who have actually bothered to learn the options and so don't confuse.

Personally I think apps should default to helping the user, but provide options for the user to take more direct control. I can understand wanting to search from the title bar, but leave in an option to turn off this functionality.

A default user should probably only have a "log out" button that eventually turns the computer off ( http://www.joelonsoftware.com/items/2006/11/21.html ), but if I know the differance between hiberante and sleep (like I do), then let me have my 7 choices.

> This is what Eclipse does, and it saves lots of time correcting simple

> typos and other innocent syntax errors. But the Eclipse compiler is so

> fast that it hardly matters. I'm sure that this could be worked into VS easily.

ReSharper does the same thing for C#. Its a godsend! I can live with the enormous amounts of memory required to run ReSharper just for this feature!

It basically just compiles the code in the background as you type, and highlights any errors that it finds. I also love the way it has a little bar down the right-hand side of the page where it puts red (for errors) and yellow (for warnings) marks to show where in the file the errors/warnings are. Just click on the mark and it highlights the corresponding line.

>ReSharper does the same thing for C#. Its a godsend! ...

>It basically just compiles the code in the background as you type, and highlights any errors that it finds.

Visual C# 2005 does this for C#, too.

Visual C++ doesn't; apparently C++ is just too confusing a language for Microsoft to do realtime compiles on.

Honestly, I don't understand why everyone has problems with the trailing semicolon. Shouldn't you be in the habit of putting a semicolon at the end of every thought? Last time I missed one was because my thought was interrupted and I went someplace else in the code before returning. Quick compile error sorted that out.

Personally, I'm very happy that people typing "thud" in Firefox get to my website and not the domain squatters who want $20,000 for thud.com.

Erzengel: The VS2005 IDE doesn't actually do a full compile in the background, though, so it doesn't pick up all errors. It DOES pick up missing or extraneous semi-colons. However, ReSharper will (for example) highlight unused local variables, unused private variables, private variables that are assigned a value but never used, etc, etc. (I know these are actually warnings, not errors, but you can see what I mean - for example the IDE won't highlight an erronous "a = b;" statement if a or b are not defined).

There's no real reason why C++ couldn't do the same thing. One of the (only, IMHO) benefits of the single-pass compiler is that it can just compile the source file you're working on, and with good pre-compilation (it could even be automatic - for the error-highlighting, it could precompile ALL referenced headers) it should be pretty quick. But yeah, C++ is a MUCH more complicated language to compile than C#. Oh well :)

"There is another possible reason that long distance calls require a 1: area codes and local exchange codes (the first three digits of a phone number) can overlap, and unlike cell phones (and URLs), landline phones have no unambiguous signal that a phone number is complete."

That doesn't apply when you always have to dial the area code, such as in Eastern Mass.

When they first switched to this, the area code was optional, for I think a year.  If you previously could dial 7 digits, then you still could, but you could also dial 10 digits, which used to be an error.  This gave people a chance to transition to the new style.  After that, the area code was required.  The presence or absence of the 1+ is still very strict.  If it's not there and it needs to be, it's an error.  If it is there and it shouldn't be, it's an error.

Google should make it so that typing an address like Yahoo.com or whatever takes you to the site, instead of search results.

And last I checked, you could put search terms in the address bar, and IE will search for them. You just need a way of making it use Google. Which I believe is simply a matter of fiddling with the settings.

Rybec:

It's a matter of IE7, works like a charm.

I once had a friend who used google to get to everything that he didn't consider "nessicary"  he set google to his home page and typed all the web addresses in there.  That way the "important" ones stayed in his recently used drop down list (I don't pretend to understand it but it worked).  He used this in place of bookmarks.

The phone system in the USA always seems rather clunky to me.

Initally in Australia we had 6 digit numbers with 3 digit area codes, except for capital cities which were a 7 digit number with a 2 digit area code

eg, my home town had numbers like 055 23xxxx

those first 2 digits would change in different areas within that area code

Numbers in Melbourne for example, were 03 xxxxxxx

But when this was going to start causing issues wiht not having enough numbers, things were changed.

Capitol cities got a digit added to the start of the number, a 9 in most places, therefore 03 9xxxxxxx and the other areas in that state got their area code stuck to the number, and the whole state was placed in the same area code. so numbers in my home town became 03 5523xxxx

Now you just dial the 8 digit number for anywhere in the same state, and the 2 digit area code for interstate calls

Easy. Your phone system sucks.

Monday, November 27, 2006 1:59 PM by Zoltran

> I hate to be offtopic...but can someone point

> me to a page that describes the 7/8/10/11

> digit dialing rules?

There are a ton of different rules and they vary by city or district or region.

Many cities had the 1+ rule for toll calls for the reason that several other commenters already stated.  Some of those cities changed their rules in recent decades but it looks like some didn't.

Some cities didn't require the 1+.  Some ignored one 1+ but recognized a second 1 (11+) and I don't know what they did in such cases.

In Canada and the US and part of the Caribbean, until a few decades ago, if the first digit was 2 through 9 and the second digit was a 0 or 1 then the total number of digits would be 10 including an area code.  Then some cities added local 7-digit numbers in which the first digit was 2 through 9 and the second digit was a 0 or 1.  So then suddenly they needed to use 1+ to distinguish whether an area code followed (total 11 digits including the 1+) or not (total 7 digits without a 1+).  So 1+ could no longer be used for its older purpose.  I thought that every city would have to change to this system, and I'm surprised to see here that some didn't have to.

In most of the world city codes start with a 0.  A phone dial could be locked so that local calls could be dialled using digits 1 through 9 but long distance calls couldn't be dialled.  Historically the caller would either pay the owner of the phone or some public phones would accept coins, and the cost of the call would be constant (up to 3 minutes anyway) because it was guaranteed to be a local call.  Then public phones became smarter, able to calculate or communicate with central office switches to get calculations of long distance call costs.  Then locks were no longer necessary, and then new local phone numbers started to get some 0 digits in them.  In most of these cases there were no rules for the number of digits in a phone number.  The switches still had to inform the phone when a complete number had been recognized.

Monday, November 27, 2006 12:38 PM by SM

> re: Tarig

>>It much easier to just type "yahoo" in the

>> search box and click the link in the search

>> results page than type in additional "www."

>> and ".com" and then press enter.

>

> Ever try ctrl-enter in the url box?  Try it --

> just type yahoo then hit ctrl-enter.  Works in

> IE and Firefox.

Yup, I hate that.  Right-click a link somewhere to copy the URL, do a Ctrl+V into IE's address bar, hit the Enter key, and notice too late that the finger came off the Ctrl key a millisecond too late so IE automatically appended a ".co.jp" to the URL.  Then instead of browsing http://x.co.jp/y/z we're getting a 404 on http://x.co.jp/y/z.co.jp

Brillant.

Dialing 1 – Old fogies like me will remember that switches used to distinguish area codes from exchange prefixes by the convention that the second digit of the former was always a 0 or 1 and the second digit of the latter was never 0 or 1.  When more area codes were needed and the convention had to be abandoned, the prefix 1 was added to signal that the following digits should be treated as an area code.  The newer convention of using the prefix 1 to signify long distance even without an area code is, IMO, completely goofy and needlessly complicated – as if you have to know who the carriers are before you can identify the destination.

Semicolon – UCSD Pascal did allow a superfluous and technically erroneous semicolon before END (by allowing empty statements within a block).  But this was only after much heated debate.  The purists succeeded in disallowing a semicolon before ELSE by claiming (falsely, I think) it would require adding unbounded look-ahead to the parser.  IMO, the definition of semicolon as statement separator, not terminator, was a little too cute.

I've tried to check out from web stores where I've been yelled at "the credit card number you entered is a MasterCard, not a Visa; please select 'Visa' and try again".

...while wiping out the card number I entered.

Yes, if you know what I meant, then please just @#$%^-ing DO THE RIGHT THING.

"I agree with kokorozashi, when Microsoft "dumbs down" the UI they convert more geeks to Linux."

Microsoft doesn't care about geeks, there are less geeks than non-geeks. If they have all the non-geeks then they win ;)

This is why Linux will never win, because all the Linux bigots think they are the ones who make an OS successful. Microsoft KNOWS it's the average user that really counts.

Here's a thought. The open source browser Epiphany ( http://www.gnome.org/projects/epiphany/ ) embraces this. It doesn't even have a search bar. If you type something that doesn't look like an URL in the URL box it will display a choice of 'Search for Whatever' in the autocomplete drop-down.

I've used that browser for a while and it never bothered me. Although I never searched Yahoo :)

PingBack from http://blog.strafenet.com/2006/11/28/the-method-never-justifies-the-product/

"If you type something that doesn't look like an URL in the URL box it will display a choice of 'Search for Whatever' in the autocomplete drop-down."

IE7 does this too.

"If you type something that doesn't look like an URL in the URL box it will display a choice of 'Search for Whatever' in the autocomplete drop-down."

IE7 does this too.

Monday, November 27, 2006 8:53 PM by Mikkin

> UCSD Pascal did allow a superfluous and

> technically erroneous semicolon before END (by

> allowing empty statements within a block).

> But this was only after much heated debate.

Oh wow, I can just imagine the heated debate.  Since the standard allows empty statements, conforming implementations (though the Pascal standard doesn't use that term for them) are required to accept programs that have superfluous and technically valid semicolons before END.  There must have been a big debate about whether it's allowable to build a nonconforming implementation that rejects valid programs, then adjust the implementation to give a false warning instead of error and thereby accept valid programs, and then, um, whatever.

> The purists succeeded in disallowing a

> semicolon before ELSE by claiming (falsely,

> I think) it would require adding unbounded

> look-ahead to the parser.

Now I think you're talking about the standard committee instead of the UCSD implementors.  The committee's lie was pretty close but not exactly that.  They asserted that ambiguity would result.  The Algol, PL/I, Pascal, and C versions can all be described using ambiguous grammars if you wish, they can all be described using unambiguous grammars if you wish, and a modified Pascal version that could accept an optional semicolon there could also be described by either an ambiguous grammar or an unambiguous grammar.  I never figured out why the Pascal committee chose that particular topic to be anal-retentive about and why they chose that particular topic to give such a dishonest reason for.  In a few other cases where they refused to yield to popular demand, at least they admitted the reason was their own sense of aesthetics.

I think things are getting a little technical when the main point was customer service.

The best part about typing your URL into the search engine is that you will generally get to the right place, rather than if you leave off the www. and some ratbastard sends you to their garbage common results page just for forgetting three little letters.

I'm quite happy with the concept of sending a url to a search engine, one thing I would ask of all search engines, however, is that if the result looks like a good URL, then return the link to that web-page as the first result regardless of how high that page would have appeared in the list of results.

The fact that most people open up their browser and type what they want into the search box on the page means that if you tell them to go to www.SomeObscureSiteThatMsnAndGoogleDontKnow.for.example.com, then people start complaining that they can't find your site ... which is somewhat annoying when your trying to build interest.

I would definitely agree with the sentiment for search engines to take you straight to a URL.

I love the idea of the search toolbar in browsers like Firefox and would love to make full use of it. However, I love my uber-slim interface. I still can't understand why the search toolbar and the url bar can't happily live in the same space. I type in a URL, you take me there. I type in something else, you search it for me. I type in a URL I want to search, I select that option from the drop-down list.

"I've tried to check out from web stores where I've been yelled at "the credit card number you entered is a MasterCard, not a Visa; please select 'Visa' and try again"."

To top that, a few years ago I tried making a credit card call via AT&T. After the prompt 'enter any major credit card number', I entered my Visa card number. No luck - and, IIRC, a rather unhelpful message that the number was 'not recognized'. Eventually, I reached an operator who told me that in AT&T's mind, Visa don't count as "major".

I outline my proposed system here:

http://icr.vox.com/library/post/urls-and-search-happily-coexisting.html

640k: It's harder than you think.

See this previous entry: http://blogs.msdn.com/oldnewthing/archive/2004/11/24/269237.aspx

BryanK,

Yeah Firefox does this as well

SM,

Not all end users know about ctrl + enter

I use google to search for "old new thing" as I can't remember the URL, I could add a bookmark, not sure why I don't really but hey. Just typing "g old new thing" in opera is easy enough.

jg:  To get to this page, I always do a google search for "old new".  You can save yourself a word of typing!

Completely agree on the phone numbers issue. Here in Ireland there are a lot of small towns that are only after moving off 5-digit telephone numbers in the last few years. Many businesses haven't updated their cards or even websites so you dial the area code and number 045-12345 and get a message saying "That number has changed. Please insert 87 before the number." It's really frustrating - you know what's wrong, fix it for me.

"Not all end users know about ctrl + enter"

I didn't until I read this post.

and one last reason.....people surf the net from work.....and seeing a google page in the browser memory is easy to explain......www.yahoo.com...or www.hotteens isnt...

lol

When you press ctrl + enter IE tries to be smart and appends your current country's TLD (you have to check your locale to be able to tell what IE appends). And there's a bug in the Swedish localized version of IE when you press ctrl + enter, it's appending the wrong TLD. This bug has survived several years & service packs and hasn't been corrected yet. Useless feature!

--One solution would be a separate program for geeks like me with a simple text box which is highly configurable but defaults to doing what I say rather than what it thinks I must mean. But of course that program won't be installed everywhere.--

It's called "QuickSilver" on Mac OS X. Unfortunately, there's no Windows port of it... it's a brilliant piece of software. Sorry for being off-topic, I just have to get in that shout-out.

> jg:  To get to this page, I always do a google search for "old new".  You can save yourself a word of typing!

To get to this page, I type in "bl", down-arrow, enter: FF remembers that the most-visited URL starting with bl was blogs.msdn.com/oldnewthing/, so it shows that first.  Four keystrokes!  ;-)

"I've tried to check out from web stores where I've been yelled at "the credit card number you entered is a MasterCard, not a Visa; please select 'Visa' and try again".

...while wiping out the card number I entered.

Yes, if you know what I meant, then please just @#$%^-ing DO THE RIGHT THING."

Web sites don't need to ask what kind of credit card you are using (at least for the common ones).  Cards that start with a 4 are Visas, cards that start with a 5

are Master Cards.  Discover cards start with 6, and I think all of them start with 6011.  I'm not sure about American Express.  

You're right, TJIC, and now I'm impressed when I use a shopping site that doesn't ask me what kind of credit card I am using.  

My slight experience doing software on a store system in a restaurant is that the verification request gets sent to one single payment processor regardless of the card you use.  They can tell the difference; why complicate things by asking the user?

To get to THIS site, I click the icon on the Links bar, which I have conveniently moved just to the right of the menu (which is still visible in IE7 if you look up the Registry hack).  

Without the menu visible, File/Send Link by Email is not (easily) doable.

IJC: "I type in a URL, you take me there. I type in something else, you search it for me. I type in a URL I want to search, I select that option from the drop-down list."

IE does that unless you select "Don't search from the address bar".  In other words, if you type a non-URL into the address bar, IE does a search for you.

While we're at it, when will someone build a browser smart enough to know that if I type "www,yahoo.com" I almost certainly meant "www.yahoo.com" ????

Articles from around the web on silly users, "broken windows", personal e-mail a

The object-based language that Borland devised for their "Paradox for Windows" IDE uses no statement separators or terminators at all.  If you code "A = B C = D" it's just two assignment statements.  (On the other hand the semicolon introduces a comment, so "A = B; C = D" is one assignment statement with very misleading documentation.)

As far as I know the lack of statement delimiters has never given anyone any trouble.

-Wang-Lo.

Regarding David Walkers comment that IE already does my much desired integration of the search bar and URL bar, I refer you to the email he very kindly (it was much apreciated) sent me:

"Yes, but it's a poor implementation. As has been stated, it often ends up searching mistyped URLs (unless you use OpenDNS). You can't easily select a search engine.  It also searches for the URL every time you try and search for it, meaning you have to wait a long time if you're searching. Not to mention the miriad of other problems with IE7s interface in the url and search bar alone. It's not the same as integrating the search BAR and the URL bar. It doesn't integrate all the features properly or well."

BryanK,

"Um, Firefox does this by default, if I understand what you're doing."

It's a bug: I habitually type the first few letters of a frequently-visited URL in Firefox's address box and hit down arrow/return. If I miss the down arrow key, I get ambushed by that misfeature. The site it sends me to is, invariably, nothing that interests me in the slightest. Which is obvious, when you think about it.

This is usually just a mild annoyance, but there was the time when I typed "alt" and missed the arrow, and it sent me to the glaringly NSFW alt.sex.stories repository. No confirmation prompt, of course; it just pulled a random site out of its fundament and sent me there. 2.0 now sends me to Google when I do that; the alt.sex.stories repository is now the fourth link down the list.

The problem that one time, was the customer standing behind me when this happened. Fortunately, he saw the humor in it. Some people wouldn't.

It's insane to imagine that your code can guess what the user wants, based on a random typographical error. And as somebody noted above, the IE version of the same bug is even more destructive.

You can theoretically disable Firefox's typo-ambush feature in about:config somewhere, but last time I tried, the setting didn't actually disable it. Oops.

640k : "This makes me think of the message explorer is showing when you drop a file on a task bar button. If it knows whats wrong, and how to do it right, why doesn't explorer do it itself?"

It knows you dropped a file on the task bar.  That is slightly different from knowing "how to do it right".  (Except in the trivial sense that dropping it somewhere other than the task bar is right).  

J. Edward Sanchez linked to a previous article explaining there is no programmatic way to hand off the drop to the program.  So that leave un-minimizing the program and "randomly" picking a section of the displayed window to do a drag & drop drop on.

That could lead to very unpredictable, and "wrong" results.

Sure, for programs like MS Word dropping a dragged file onto an open copy of the program means 'open file'.  But it doesn't have to.  Some programs may only support file dropping in specific areas of the window (I recall some older shareware programs that would let you drag and drop a file onto a text entry box and the box would be filled out with the full path to that file; dropping the file anywhere else would be ignored).

Or a program might support multiple drag and drop sites.  Conceivably you could have a program with one drop area for secure copy to default remote computer, one for secure copy + secure delete, and a third for secure delete. (Contrived example sure; but you hate to see explorer randomly pick a pixel of that app to drop the file on).

Less contrived, the application might have a windows modal file dialog box open, dropping the file in the middle of that would copy (or move; depending on explorer) the file from its original location to the displayed directory.  Almost certainly a "wrong thing".

Phill, that happened here too.  It was how the phone company told you that what you're doing is wrong, and how to fix it, before they make it just not work at all, and reuse that space for something else.

David, AmEx starts with 3, and is 15 digits instead of 16 digits.

"Now I think you're talking about the standard committee instead of the UCSD implementors."

I see I was unclear Norman.  I did not mean to mention anyone’s particular compiler.  I was talking about the good folks in the UCSD Department of Applied Physics and Information Science who, under Kenneth Bowles, adopted the null statement hack (and other innovations) in the hope that it would become an accepted standard.  The null statement is indeed conformant UCSD Pascal.  It was not generally accepted (Wirth 1974) Pascal at the time (ISO 7185:1983 was a bit later), although I think some compilers already got away with allowing it because it had the virtue of correctly handling conformant source code.

The look-ahead red herring was not because of an ambiguous context-free grammar, but because a really simple (not to say minimalist) recursive descent parser would pop the IF before seeing the ELSE part.

Raymond, my recollection of the paraphrase (from UCSD’s UCSD Pascal compiler circa 1975) is “Unexpected identifier, possibly missing or extra semicolon.”  The error in the message is that ELSE is not an identifier.  Sorry for wandering off topic.  Aren’t you glad you brought it up?

I once tried to make a call and got the message "You must dial a 1 or 0 in front of this number."  I then dialed the 1 and got "It is not necessary to dial a 1 or 0 before this number."

Back on topic - Typing an address in a search box is natural and convenient.  That the user is using a tool with larger scope than necessary should not surprise or bother anyone.

On the other hand, spontaneously launching a search when an address is not found is just evil.  Yes, someone who knows how can turn it off, but we were considering unsophisticated users here.  It would be helpful and noninvasive to pop an offer to search.  It would be convenient to put a "search" button next to the "Go To" button on the toolbar.  But, like the labels used in the snail post:  "Do not forward, return to sender only."

> Mikkin: It would be helpful and noninvasive to pop an offer to search.

I agree, however if such an option existed the ideal case would be to have it configurable options such as "show search popup prompt" "automatically search default search engine" "automatically search search engine of user choice"

I think that there will always be people who love "intelligent" DWIM systems, and those who hate them. Me, I tend to loathe them with a fury I'd hitherto reserved only for politicians.

I was surprised to find myself coming to love the Firefox address-bar's behaviour. Being able to type "<companyname> annual report 2006", "old new", "<sitename> sitemap", "<comicname> archive"... these things have become second nature to me, and I would not have them removed. They save me time and effort, compared to using the search bar. It becomes intuitive which queries will return the  correct result and which will not.

Like tabbed browsing, it is a discovery that has changed the way I use the browser.

The address bar is now for when I want to go to a specific address on the net, which I know exists, and I can describe it well, by URL or by phrase. It's like jumping in a taxi, telling the driver where I want to go, and being taken there.

The search bar is for when I want a range of opinions on a topic, or when I cannot describe the page well and want to be served a palette of options that may be it. It's like consulting a directory to find the address of the place I want to go.

Both have their place, but they are different, and shouldn't, I feel, be merged into one "DWIM" field. One field with two buttons, though, yeah.

"Phill, that happened here too.  It was how the phone company told you that what you're doing is wrong, and how to fix it, before they make it just not work at all, and reuse that space for something else."

But they're not going to use that number for anything else. The numbers are now 7 digits long, not 5.

I think it's just that the phone company doesn't give a damn about backwards compatibility and to hell with all the companies that spent money getting business cards printed up!

Phill, those 5 digits could be used in the future as a prefix of a 7 digit number.  No complete phone number can be a prefix of another numbe