Alan Kay's EuroPython Keynote

> Please don't read this message if you don't have time or inclination
> for a quasi-philosophical ramble down Smalltalk Lane. I'm posting
> this here to share some experiences I doubt are new or unique but
> which I haven't seen discussed here in the many months I've been back
> on the list.
>
> I have been involved in Smalltalk on and off for a good many years.
> It keeps calling me back, like the Sirens, whenever I go in search of
> a new tool because the one I'm engaged in at that moment falls short
> or disappoints or just annoys me. As a result of this on-again, off-
> again love affair with Smalltalk and Squeak, I'm far from as
> proficient a coder or knowledgeable a designer as I would certainly
> be by now if I'd stayed put here. But I haven't been able to do that,
> for a host of reasons that are mostly boring and unique so I won't
> relate them here.
>
> Today I had a conversation with a colleague and friend I had referred
> to Squeak as a possible solution to a specific set of problems he is
> working on for a client. He spent a full day exploring Squeak and he
> came back with an observation that I found difficult to answer.
> "Why," he asked me in all sincerity, "is Squeak so ugly? Smalltalk
> has been around 30 years. It's been in the hands of great design
> firms like Apple and Disney. It's had IBM backing. Doesn't anyone in
> the Squeak community understand how a polished, modern user interface
> would help to sell their technology? Other than wxSqueak, which seems
> basically moribund [I disabused him of this notion in our
> conversation, but that was his finding on his own], there's nobody
> out there talking, thinking or working on a professional-looking UI
> for Squeak's IDE or for deployment of applications! What's going on
> there?"
>
> So I spent a couple of hours looking at the question he raised and
> what I *think* I learned is that because of the way Smalltalk
> implements graphics at some deep level beyond my ability to penetrate
> the image, modifying its basic UI to use a more modern and reactive
> user experience would be a major, major challenge. After 30+ years,
> there is no way to do native UI widgets (other than wxSqueak if and
> when it gets finalized and hopefully incorporated) let alone custom
> widgets that look polished and professional. I was able to determine
> that there appears to be a class (PNGReadWriter) that would
> facilitate the import of PNG images, e.g., to use as controls. With
> enough time and understanding, I could presumably figure out how to
> import a graphic (PNG or other) and make it behave like a button, but
> then getting it into an app layout would require another level of
> understanding.
>
> You get the idea. (And please don't spend time telling me how to do
> that particular task; I don't have the expertise or interest anyway.
> It was merely an illustration of what the problem appears to be.)
>
> I explained to my friend that Squeak has been used primarily for
> research and education, not for the creation and deployment of
> commercial applications where a standardized platform-specific UI was
> important. For him, that's a reason to avoid Squeak altogether.
>
> So with that (probably overly long) background, I can ask my
> question, on my friend's behalf.
>
> Why, after 30 years, does Squeak still appear to be a non-standard,
> almost toy-like user experience in the IDE? Is it the case that
> changing that would be far too complex to undertake? Or is it that
> the community of Squeak users just isn't largely motivated to worry
> about this subject? Or is the absence of an economic incentive the
> problem? Or IS there a problem?
>
> Thanks for any wisdom you can share. This is one of the two big
> objections I *always* get when I recommend someone look at Squeak as
> a possible solution to a problem for which it appears to me to be
> ideally suited linguistically and architecturally.
>
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> Dan Shafer
> Technology Visionary - Technology Assessment - Documentation
> "Looking at technology from every angle"

> Your original question mentioned IDE's Dan:
>
>>>> almost toy-like user experience in the IDE? Is it the case that
>
> therefore I thought the audience you were referring to were  
> developers.
>  For developers I am less-tolerant of their whining about look (but  
> not
> feel, because Nicolas Cellier's points are excellent).
>
> But since you have clarified you apparently are talking about
> *end-users*, I'd like to clarify, I think, for end-users look is much
> more important than for developers.
>
> But I doubt native/standard/whatever look alone will be enough  
> because,
> as I mentioned, IBM's VisualAge Smalltalk had native widgets PLUS tons
> of interfaces to major business systems (even CICS!) that no Smalltalk
> will probably ever have again.
>
> I think Cees nailed it, its so hard to overcome the anti-marketing
> forces.  IBM was selling their VA Java product for half the price of
> their VA Smalltalk.  Its amazing VAST lasted as long as it did.  I'm
> sure they were gratified (again) to sell the small interim-step in
> technlology (VAJ) as a huge wave of revenue.  But I think they made a
> big mistake, now where are they gonna go?
>
> Better looks will never hurt though..
>
>
> --- Dan Shafer <[hidden email]> wrote:
>
>>
>> On Jul 7, 2006, at 12:50 PM, Chris Muller wrote:
>>
>>>> Why, after 30 years, does Squeak still appear to be a
>> non-standard,
>>>> almost toy-like user experience in the IDE? Is it the case that
>>>> changing that would be far too complex to undertake? Or is it that
>>>> the community of Squeak users just isn't largely motivated to
>> worry
>>>> about this subject? Or is the absence of an economic incentive the
>>>> problem? Or IS there a problem?
>>>
>>> 1.  What does that really mean, looks toy-like?  What's it gonna
>> take,
>>> please tell us?  A manly "gray metal" look like we've seen before
>> or
>>> some more rainbow gradients like we've seen before?
>>
>> I can see I touched a sore spot here. First, please remember I'm
>> passing on reactions from others, not expressing my own views here.
>> I'm trying to figure out how to make it possible for me to get
>> projects and contracts using Squeak from people with pre-conceived
>> ideas about what an application looks like.
>>
>> Short of true native platform widgets, I think it's safe to say that
>> the vast, vast majority of computer users are accustomed to certain
>> kinds of shapes for certain kinds of objects and that when we vary
>> from those standards, we'd better have a really good reason for
>> asking them to pay attention to our interface rather than to the task
>> they are trying to accomplish. I don't know if that's "manly" gray
>> metal or rainbow gradients or just round-cornered rectangles with
>> default buttons flagged somehow. The point is that it is NOT the UI
>> one finds in Squeak today.
>>
>>> 2.  What is THE "standard"?  Microsoft?  What, then, when Microsoft
>>> changes its look again?  Are they then "non-standard" or their
>>> followers?
>>
>> Again, I think it's less an issue of choosing a standard than it is
>> of choosing a UI experience that maps to the user's established
>> expectations.
>>
>>> 2b.  Do you remember this rubbish, "All windows programs look alike
>>> therefore once you've learned one you've learned them all.."?
>>
>> Yeah, only I heard it said about Mac programs. Same rubbish, though.
>>
>>> 3.  Even if it can be "offically" labelled "toy like," what is
>> wrong
>>> with that?  Too wimpy-looking?  What's wrong with wimpy-looking as
>>> long
>>> as its easy and functional?
>>
>> Sorry to keep beating the dead horse here, but the issue is that it
>> is toy-like *compared to user expectations* set up by other programs.
>> I find the EToys UI refreshing. My potential clients and colleagues
>> whom I wish to interest in Squeak just don't.
>>
>>> 4.  Speaking of wimpy, someone (not you) once suggested "Squeak"
>> was a
>>> wimpy-sounding name, what do you think of "KA-POW!"?
>>>
>> LOL. I think Squeak is a weird name for a programming language. But
>> not weirder than C or Java or Ruby.
>>
>>>> Thanks for any wisdom you can share. This is one of the two big
>>>> objections I *always* get when I recommend someone look at Squeak
>> as
>>>> a possible solution to a problem for which it appears to me to be
>>>> ideally suited linguistically and architecturally.
>>>
>>> So I guess this group sets its priorities to building something, as
>>> you
>>> said, "ideally suited linguistically and architecturally".  They
>>> haven't spent as much time playing endlessly with colors, shadows,
>>> gradients, only to please the latest group of popularist
>>> superficialites who'll be gone as soon as the next Wired article
>> tells
>>> them where they need to go next to be cool.  Whew, sorry to say
>> that,
>>> at least you have my honesty.
>>>
>> I am at least somewhat sympathetic to your thinking here, Chris. All
>> I'm really trying to do is find answers that will satisfy and/or make
>> sense to the great majority of people who've had their user
>> experience expectations shaped by the market.
>>
>> Dan
>>
>>
>>
>
>

>   Nicolas,
>
> > Apart lack of keyboard shortcuts (at least standard ones) that are
> > breaking productivity, one thing that would mostly afraid someone having
> > classical end user applications in mind is the halo handles... What are
> > these handles usefull for end users? Are they just there so the end user
> > can break the UI? I perfectly understand negative reactions.
>
>   Of course these handles are useful for end users who are doing
> eToys^^;
>
>   Other than that, there is a package called "Lockdown" by Ned.  This
> may not work with latest Squeak versions, but you'll see the idea.
>
> http://map.squeak.org/package/0c6d3bda-adbb-496a-ae84-8d4c145d14b5
>
> -- Yoshiki

> Please don't read this message if you don't have time or  
> inclination for a quasi-philosophical ramble down Smalltalk Lane.  
> I'm posting this here to share some experiences I doubt are new or  
> unique but which I haven't seen discussed here in the many months  
> I've been back on the list.
>
> I have been involved in Smalltalk on and off for a good many years.  
> It keeps calling me back, like the Sirens, whenever I go in search  
> of a new tool because the one I'm engaged in at that moment falls  
> short or disappoints or just annoys me. As a result of this on-
> again, off-again love affair with Smalltalk and Squeak, I'm far  
> from as proficient a coder or knowledgeable a designer as I would  
> certainly be by now if I'd stayed put here. But I haven't been able  
> to do that, for a host of reasons that are mostly boring and unique  
> so I won't relate them here.
>
> Today I had a conversation with a colleague and friend I had  
> referred to Squeak as a possible solution to a specific set of  
> problems he is working on for a client. He spent a full day  
> exploring Squeak and he came back with an observation that I found  
> difficult to answer. "Why," he asked me in all sincerity, "is  
> Squeak so ugly? Smalltalk has been around 30 years. It's been in  
> the hands of great design firms like Apple and Disney. It's had IBM  
> backing. Doesn't anyone in the Squeak community understand how a  
> polished, modern user interface would help to sell their  
> technology? Other than wxSqueak, which seems basically moribund [I  
> disabused him of this notion in our conversation, but that was his  
> finding on his own], there's nobody out there talking, thinking or  
> working on a professional-looking UI for Squeak's IDE or for  
> deployment of applications! What's going on there?"

> So I spent a couple of hours looking at the question he raised and  
> what I *think* I learned is that because of the way Smalltalk  
> implements graphics at some deep level beyond my ability to  
> penetrate the image, modifying its basic UI to use a more modern  
> and reactive user experience would be a major, major challenge.  
> After 30+ years, there is no way to do native UI widgets (other  
> than wxSqueak if and when it gets finalized and hopefully  
> incorporated) let alone custom widgets that look polished and  
> professional. I was able to determine that there appears to be a  
> class (PNGReadWriter) that would facilitate the import of PNG  
> images, e.g., to use as controls. With enough time and  
> understanding, I could presumably figure out how to import a  
> graphic (PNG or other) and make it behave like a button, but then  
> getting it into an app layout would require another level of  
> understanding.

> You get the idea. (And please don't spend time telling me how to do  
> that particular task; I don't have the expertise or interest  
> anyway. It was merely an illustration of what the problem appears  
> to be.)
>
> I explained to my friend that Squeak has been used primarily for  
> research and education, not for the creation and deployment of  
> commercial applications where a standardized platform-specific UI  
> was important. For him, that's a reason to avoid Squeak altogether.
>
> So with that (probably overly long) background, I can ask my  
> question, on my friend's behalf.
>
> Why, after 30 years, does Squeak still appear to be a non-standard,  
> almost toy-like user experience in the IDE? Is it the case that  
> changing that would be far too complex to undertake? Or is it that  
> the community of Squeak users just isn't largely motivated to worry  
> about this subject? Or is the absence of an economic incentive the  
> problem? Or IS there a problem?

> Thanks for any wisdom you can share. This is one of the two big  
> objections I *always* get when I recommend someone look at Squeak  
> as a possible solution to a problem for which it appears to me to  
> be ideally suited linguistically and architecturally.
>
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> Dan Shafer
> Technology Visionary - Technology Assessment - Documentation
> "Looking at technology from every angle"
>
>
>
>

>
> On 7-Jul-06, at 2:45 PM, Daniel Poon wrote:
>
>> +1
>>
>> UI widgets are a solved problem. It is part of the operating system.
>> Like writing a file system is done, writing UI widgets is done too.
>
> Now that I have to disagree with. There is no OS/GUI /platform that I
> have so much as heard rumours of that qualify as having 'solved UI
> widgets'. They are all sufficiently steeped in effluvium as to smell bad.
>
> I don't think filing systems are 'done ' either; certainly microsoft
> hasn't managed to finish that damn big file system it's been promising
> RSN since I first moved to the US in 91. Unix is not exactly a shining
> example either.
>
> Just because lots (all?) of the current commercial platforms do
> something  doesn't mean it is a good - or even acceptable - idea.

> INteresting perspective, Trygve, and one that appears to have held  
> sway in the world of Smalltalk for a long, long time.
>
> "Don't let the paint dry just yet, we're still inventing!" But  
> THIRTY YEARS???!!!
>
> A few people have suggested, on this list and off, that I refer my  
> friends and colleagues to Dolphin Smalltalk (and, on OSX, to Ambrai  
> Smalltalk, which isn't nearly as far along yet). But one of the  
> reasons Smalltalk is in contention as a language for these projects  
> is its cross-platform nature. I don't want to give that up.
>
> I've been around the fringes of this stuff for more years than I'd  
> like to admit. I've written several books on it. I've helped write  
> two major systems in it. I'm kindly disposed toward it. AND, I know  
> what a huge undertaking it would be to fork off a variant that  
> would address the specific issues I think need addressing and an  
> even bigger job trying then to maintain compatibility with the  
> forward movement of Squeak. Yes, I know I could even build a  
> specialized vocabulary/language on top of Squeak and completely  
> alter the UI. I know others have done that. But my friends and  
> colleagues and clients have problems today that need solutions  
> today. It is just so frustrating to me that after so many years,  
> Saueak remains a laboratory, an exploration, a learning experience.  
> I am too old to start over!
>
> Dan
>
> On Jul 8, 2006, at 1:06 AM, Trygve Reenskaug wrote:
>
>> I see Squeak as a wonderful laboratory where hundreds of  
>> professional and amateur researchers can explore, share and test  
>> new ideas. My nightmare is that it should become a mainstream  
>> professional IDE prematurely. The need for "backward  
>> compatibility" could then be overwhelming, effectively hindering  
>> innovation. Send the application developers to Dolphin for the  
>> time being and permit the next Squeak version be incomatible with  
>> the last.
>
>

> On 7/8/06, Dan Shafer <[hidden email]> wrote:
>> That may be the ultimate
>> solution for a lot of what i want to do on the desktop, but for the
>> near term, that won't satisfy my clients and colleagues because the
>> wxSqueak stuff isn't yet ready to release.
>>
> Why not? Did you test it? Find issues?
>
> I used it for a project last year, and whatever problems we had, they
> weren't wx-related...
>

> Cees.....
>
> I am going by what Rob said about it. He advised me not to rely on it  
> for a project just yet, but I think he's actively working on it again.
>
> Your news is encouraging. Perhaps Rob is being overly cautious.
>
> Dan
>
> On Jul 8, 2006, at 2:59 PM, Cees De Groot wrote:
>
>> On 7/8/06, Dan Shafer <[hidden email]> wrote:
>>
>>> That may be the ultimate
>>> solution for a lot of what i want to do on the desktop, but for the
>>> near term, that won't satisfy my clients and colleagues because the
>>> wxSqueak stuff isn't yet ready to release.
>>>
>> Why not? Did you test it? Find issues?
>>
>> I used it for a project last year, and whatever problems we had, they
>> weren't wx-related...
>>
>
>
>

> Dan,
>
> My 0.4 release was reasonably solid for Win32, but the Mac version  
> wasn't as good (partly my fault, partly the state of wxMac 2.5),  
> and the Linux version wasn't really usable. I'm hoping that the 0.5  
> release (coming RSN) will work well on all three platforms.
>
> As for all of the other discussion about wx/gtk/Spoon, I liked  
> Cees' response the best: Why one UI? There are pros and cons to any  
> UI toolkit, so the more options there are for Squeak, the more  
> opportunities there will be to use it. My goal with wxSqueak has  
> been to  keep it compatible with the latest release of Squeak, with  
> minimal VM changes, and maybe eventually no VM changes at all. And  
> ultimately, it should work fine with Spoon. But I think it should  
> always be an add-on package to some Squeak/Spoon core.
>
> .. Rob
>
> Dan Shafer wrote:
>> Cees.....
>> I am going by what Rob said about it. He advised me not to rely on  
>> it  for a project just yet, but I think he's actively working on  
>> it again.
>> Your news is encouraging. Perhaps Rob is being overly cautious.
>> Dan
>> On Jul 8, 2006, at 2:59 PM, Cees De Groot wrote:
>>> On 7/8/06, Dan Shafer <[hidden email]> wrote:
>>>
>>>> That may be the ultimate
>>>> solution for a lot of what i want to do on the desktop, but for the
>>>> near term, that won't satisfy my clients and colleagues because the
>>>> wxSqueak stuff isn't yet ready to release.
>>>>
>>> Why not? Did you test it? Find issues?
>>>
>>> I used it for a project last year, and whatever problems we had,  
>>> they
>>> weren't wx-related...
>>>
>
>