Re: Please explain the adversion to WPF?
jwayt
to make my code "English-like" with labels that are legible, parsable
and meaningful. I respect that the meanings of rigorously defined
symbols shall not possess arbitrary extensibility. Not in a computer
program, not in an artificial syntax.
In 30 years and over 8 million lines of code where single characters
make vast differences between success and failure, I learned it's very
bad practice to hide large differences among small distinctions. In what
trouble shooting scenario would you want to look for a subtle case
distinction in "Park" and "park", much less buried in languages that
support 255-byte labels? Case distinctions measurably double that search
space. When would you WANT to convey absolute differences in object code
not readily visible in source code? Do you really appreciate code that
uses that same string for verb, subject and object?
We struggle to achieve such clarity in our code, both during composition
and beyond 72-hour short-term memory. It's hard enough to say what we
mean.
In article <op.vigy1...@bhm-is-td-l1.corporate.local>,
Il...@gmail.com says...
>
> From what I understand Troy got banned for naming those with what he felt
> were bad ideas and asking them to defend their thoughts. That's his deal,
> not mine. We just end up on the same projects from time to time.
>
> For the compiler question, my opinion is yes. If we didn't use case in
> normal writing I would say no. But we use it to relay additional meaning
> to things that the compiler cannot understand. Like "park" and "Park". The
> capital signifies proper name and distinguishes it from the verb. Look at
> all the ways I can use it http://dictionary.reference.com/browse/park to
> relay meaning. It might actually be easier on the compiler to just treat
> it verbose.
>
> Another reason I say yes is for the same reason we mix word meanings in
> english, options. It provides us with more options and ways to name
> things. I would think this makes for better readability if done right.
>
> If it were available and we were using a smart editor, it would only
> matter during definition anyway. Editor figures out the right one to use
> based on type and such. Well, VS and several others do.
>
> Ilus
>
>
>
> On Fri, 03 Sep 2010 10:58:39 -0500, Dave Fish [Sybase]
> <n0Spam...@sybase.com> wrote:
>
> > Troy, er, Ilus I agree with what you are saying about readability but
> > I question the need for case sensitivity when it comes to the compiler
> > itself. Should myVar, myvar, and MYVAR all represent three different
> > things?
Mark Maslow
news:im1386dorhg0brb1j...@4ax.com...
> "All" the DataWindows need to do is speak AtomPub and HTTP (see
> http://odata.org). Just as the DataWindow generates say an SQL
> UPDATE Statement today, it can generate a PUT HTTP request with the
> appropriate XML payload - there's a one-to-one correspondence from DW
> row/column status, just as with SQL - it's all CRUD after all.
>
An interesting idea. I like the concept a lot, having just been
experimenting with RESTful web services using an implementation of the
JAX-RS API. We're looking into building some common web services that can
support interactions with external systems as well the web apps that we host
internally. I really like the simplicity and elegance of REST.
In my dreams, I would take it further. If PB apps could easily be made
clients to RESTful web services, it would no longer be necessary to deploy
our Java services to EAServer in order make them accessible to our PB
clients - they could just be regular web apps hosted in Tomcat. You could
also do mashups and a lot of the cool things that the Ruby guys do with such
ease.
Of course, the usual problem comes up immediately - the inability of the
datawindow to deal with complex data structures puts a rather severe
limitation on the kinds of web services you can interact with without doing
a whole lot of low level coding.
Mark Maslow
news:4c83d874@forums-1-dub...
> This is all I get in FF - just a response dialog. It does not matter if I
> select Subscribe or Cancel - all I end up with is a blank tab page (see
> attached). Whereas in IE I get the web page text and the subscription
> link.
>
You now have a bookmark entitled "Titles". Click on it to see the feed.
Chris Pollach
"Mark Maslow" <mark....@sierraclub.org> wrote in message
news:4c83ce35$1@forums-1-dub...
> "Chris Pollach" <cpol...@travel-net.com> wrote in message
> news:4c83aace@forums-1-dub...
>>I am running the latest FF on W7 and it fails to render the page. Works OK
>>though on my IE8.
>
> As Bruce said, it's a data feed. The browser will render it in whatever
> way it is configured to do so.
>
> If you install the Sage add-on, FF will show you all the data in a very
> nicely formatted page.
>
Mark Maslow
jwayt
distinctions beyond a matter of style. Ever play the game where words of
colors are printed in disagreeing colored ink and you have to say the
color of the ink, not the word? It's a similar problem.
Consider Japanese versus Chinese characters. Japanese Katakana has a
limited set closer in number to US ASCII, whereas the Chinese use over
1000 characters. This allowed Japan to adopt computer technology
significantly faster than China. We were able to load Katakana on dumb
terminals in the early 1980s while no Chinese sets were available.
Similarly, I prefer to track 26 letters to distinguish labels rather
than 52 because of the double burden. It's bad enough dealing with
numbers that look like letters, i1lI, O0o, etc., where they're not side-
by-side to contrast.
Have you ever tried to encode using base64 to number something? It's an
easy job for the computer, but a burden to transcribe accurately for
humans. How long does it take you to type accurately one of those
"strong", base64-like passwords they assign for you?
I'm a barcode technologist and can tell you the typo error rate for
numbers is at least 20 times lower than alphabet. It's not by accident!
My preference for case INsensitivity is not arbitrary and personal, but
grounded in 30 years of data entry and data collection, programming in
many languages. I've seen where mistakes occur in others and myself. My
error rates are just like theirs.
I've seen tons of code and produced tons myself. Good and bad.
Case pickiness introduces many problems in distinctions without obvious
differences. I am challenged to think of one problem it solves that
style without case-enforcement couldn't. Crap code doesn't become gold
in the presence of case-enforcement.
In article <4c728758.1e1...@sybase.com>, Big Jim says...
>
> No one said it solves all coding problems, but it's a start.
> Chris, maybe for you they could go back to something similar
> to VFP whereby you didn't need to instatiate a variable,
> just type any old thing in and let the compiler think it is
> a variable.
>
> > B.S.
> >
> > I have seen absolute crap C, C++ and C# programming!
> >
> >
> > "Big Jim" wrote in message
> > > news:4c72822b.1d4...@sybase.com... If you
> > > switch to a case insensitive language such as C#, you
> > > will no longer notice after a couple of weeks working
> > > with it. Case insensitive languages are only liked by
> > > lazy programmers. They promote bad coding habits!
> > >> FWIW: Case "Insensitivity" is what most DBA's install
> > >> their DBMS's as - so why not the development tools? It
> > >> makes more sense to me to stay the way we are in PB
> > >> classic. I hate compilers that cannot tell me if
> > >> liLoopCount is already used as liloopcount. That is
> > why >> many PB developers love it compared to Java and
> > C#. >> Business developers need "easy to use" features as
> > we are >> too busy trying to meet business deadlines than
> > debugging >> case / typo problems IMHO!
jwayt
says...
>
> Thanks very much Brad.
>
> Maybe it's time for a poll:
I vote 2
> Vote "2"
> - Silverlight support is useless to me unless I can migrate existing
> applications (with refactoring on the developers' part).
My reasoning is I can refactor better once the code is migrated. I have
working migrated code that I can move to NVO's where appropriate within
the same IDE. Discovery is enhanced, opening referenced definitions,
etc.
Tyler Cruse
displaying function/window/object names. This is bad in every case.
2) Since most of us build solutions that involve more than just PB the idea
of becoming more aligned with other tools makes some sense and adds value to
PB. Obviously, the Sybase developer believed in case sensitivity in the
runtime assemblies and the generated c# code. Also, XAML is the major part
of WPF and it is not case insensitive.
jwayt
> At the time, the business model was to sell hardware and give the
software
> away free. A 60 meg hard drive was clothes washer size and cost $30K ( +
> monthly hardware support charges )
In the late 80s I ran a 100-user system on 2 of those for the largest
office supply dealer in Houston. Now my cell phone carries 8 times that
on a chip the size of my pinkie fingernail!
Bruce Armstrong
1.5 (Somewhere between 1 and 2)
I don't believe that RIA applications will suffer from the same problems as the
current WebForm applications, in that they won't need to round trip client
interaction.
However, I do think that one of the major pieces we'd need to have implemented
in order to make RIA a success is a smooth transition to RIA Services for
distributed data access
(http://msdn.microsoft.com/en-us/library/ee707344%28VS.91%29.aspx).
The DataWindow already supports web services for select and update operations,
but it would need to be modified to
a. Only call one RIA service for insert/update/delete, not three seperate
ones.
b. Would need to generate the information for the update in the format
that the RIA Service requires. I believer that may be a diffgram, but I'm not
particularly fluent in RIA Services yet.
WCF services are already planned for a later version of PB.Net. What I'd like
to see is an RIA Service creation wizard (RIA is, AFAIK, a type of WCF Service).
This wizard could take a DataWindow as input and create a wrapper for it that
exposes the select and update RIA service methods.
I'd like to be able to use the same DataWindow for the client and the basis for
the RIA service. So I'm thinking that there would actually be two sets of
update specs. One for when it's used in the client, in which case it points
back to the RIA service. The second is when it's used in the RIA service, in
which case it points back to the database through the SQL update or Stored
Procudure.
Just a thought anyway.
On 26 Aug 2010 03:32:42 -0700,
in sybase.public.powerbuilder.futures.discussion
John Strano[Sybase] <nichtspa...@johnstrano.com> wrote:
>Thanks very much Brad.
>
>Maybe it's time for a poll:
>
>Vote "1"
>- Give us only the ability to build Silverlight applications from scratch.
>(Migrate/import previous version DWs/NVOs.)
>
>Vote "2"
>- Silverlight support is useless to me unless I can migrate existing
>applications (with refactoring on the developers' part).
>
>--
>John Strano - Sybase Technology Evangelist
>
>
>"Brad Wery [TeamSybase]" <brad...@gmail.com> wrote in message
>news:4c755e79$1@forums-1-dub...
>> Honestly, it's my strong belief that there should not be a migration path
>> to Silverlight. Silverlight targets should be reserved for new development
>> only. The reason I say this is because of the "significant paradigm shift"
>> that you mentioned. What's going to happen is that PB will end up doing
>> things totally wrong because they have to consider migrated applications
>> (similar to how user objects and visual inheritance is handled in WPF).
>>
>> PB Silverlight should be pure. If people want to make use of existing
>> code, they can turn them into assemblies.
>>
>>
>>
>> On 25/08/2010 11:29 AM, Jim O'Neil wrote:
>>> Not to argue the main point, but how many PB applications out there
>>> are 'ready' for the move to Silverlight?
>>>
>>> How many realize that Silverlight isn't exactly a web technology! Do
>>> they realize it's a plug-in, requiring adminstrative rights on the
>>> client machine? kind of like, say, the Windows and DataWindow Plug-in
>>> that have been around since PB 6.x
>>>
>>> I think PB 15 is way too late as well, but if Silverlight were
>>> available today, I dare say a lot of PB application developers
>>> wouldn't be ready.... for the architectural differences: services,
>>> asynchronicity, sandbox, UI paradigms, etc. There's been a number of
>>> various 'web technologies' in PowerBuilder for the last decade or so,
>>> why is Silverlight somehow going to be the silver (ahem) bullet?
>>>
>>> Designing applications for the 'web' is a significant paradigm shift,
>>> like client-server was from green-screen, and IMHO the reason all of
>>> the other PB web attempts 'fail' is because they don't take that
>>> fundamental paradigm shift into account. It's a noble idea to to have
>>> a magic push-button solution, but it's not reality for other than
>>> fairly trivial examples with accomodating users.
>>>
>>> I would say the best thing for the Silverlight-eager people to do now
>>> is design for WPF (after all the UI technology is the same) and do so
>>> with modern architectural patterns - services, MVVM (aka MVP), etc. If
>>> you sit on your hands waiting for Silverilght in PB 15 and expect the
>>> magic, you're going to be disappointed.
>>>
>>> Here's a reference, for example, that may be of use as you build WPF
>>> applications with an eye on Silverlight in the future:
>>> http://wpfslguidance.codeplex.com/. Yakov is also doing yeoman's work
>>> is demonstrating the refactoring and other techniques to ready
>>> yoursefl for WPF... do that now, not later, and you'll be in a much
>>> better position whenever Silverlight support in PB does come down the
>>> pike.
>>>
>>>
>
>
Ilus
were bad ideas and asking them to defend their thoughts. That's his deal,
not mine. We just end up on the same projects from time to time.
For the compiler question, my opinion is yes. If we didn't use case in
normal writing I would say no. But we use it to relay additional meaning
to things that the compiler cannot understand. Like "park" and "Park". The
capital signifies proper name and distinguishes it from the verb. Look at
all the ways I can use it http://dictionary.reference.com/browse/park to
relay meaning. It might actually be easier on the compiler to just treat
it verbose.
Another reason I say yes is for the same reason we mix word meanings in
english, options. It provides us with more options and ways to name
things. I would think this makes for better readability if done right.
If it were available and we were using a smart editor, it would only
matter during definition anyway. Editor figures out the right one to use
based on type and such. Well, VS and several others do.
Ilus
On Fri, 03 Sep 2010 10:58:39 -0500, Dave Fish [Sybase]
<n0Spam...@sybase.com> wrote:
> Troy, er, Ilus I agree with what you are saying about readability but
> I question the need for case sensitivity when it comes to the compiler
> itself. Should myVar, myvar, and MYVAR all represent three different
> things?
>
> I do agree that smart editors do make a big difference and the issue
> almost becomes moot when you use them.
>
> Regards,
> Dave Fish
> Sybase
>
> PowerBuilder Blog:
> http://blogs.sybase.com/powerbuilder/
>
> On 2 Sep 2010 13:44:17 -0700, Ilus <Il...@gmail.com> wrote:
>
>> Yet your post has perfect casing and punctuation. Your mind knows how to
>> do it. It has just been incorrectly conditioned to think it doesn't have
>> to in code. The real reason I use case sensitivity is because I prefer
>> camel casing instead of useless underscores in labeling. Why people feel
>> the need to shove underscores in everything has always been beyond me.
>> It
>> is an entirely pointless extra character.
>>
>> In today's world of smart editors, things like underscores and case
>> sensitivity are not need and need not be missed. For example, you get
>> indications if you use the wrong ones or they are automatically
>> corrected.
>> Coding standards have changed due to this. Look at a property in C#. You
>> should define the private variable right before the property and other
>> variables right before you use them. As tools change, we need to
>> leverage
>> those changes. As processes change, we need to adapt. Even though we
>> argue
>> about this sort of thing, modern editors help to keep the mistakes you
>> mention to a minimum.
>>
>> Same goes for all the people who still shove str_ or int_ or p_ or
>> whatever in front of variable names. All that shows is that your coding
>> style is out of date. Now I have seen valid arguments where people
>> preface
>> all variables with v_ regardless of type so that they sort together in
>> intellisense. I'm still trying to come to terms with that one :)
>>
>>
>>> I'm a barcode technologist and can tell you the typo error rate for
>>> numbers is at least 20 times lower than alphabet. It's not by accident!
>>
>> By that logic we should all code in binary. Only two symbols :)
>>
>> Ilus
Dave Fish [Sybase]
Regards,
Dave Fish
Sybase
PowerBuilder Blog:
http://blogs.sybase.com/powerbuilder/
>> I'm a barcode technologist and can tell you the typo error rate for
>> numbers is at least 20 times lower than alphabet. It's not by accident!
>
>By that logic we should all code in binary. Only two symbols :)
>
>Ilus
>
>On Sun, 29 Aug 2010 19:30:15 -0500, jwayt <jwayt_...@iobar.com> wrote:
>
Ilus
do it. It has just been incorrectly conditioned to think it doesn't have
to in code. The real reason I use case sensitivity is because I prefer
camel casing instead of useless underscores in labeling. Why people feel
the need to shove underscores in everything has always been beyond me. It
is an entirely pointless extra character.
In today's world of smart editors, things like underscores and case
sensitivity are not need and need not be missed. For example, you get
indications if you use the wrong ones or they are automatically corrected.
Coding standards have changed due to this. Look at a property in C#. You
should define the private variable right before the property and other
variables right before you use them. As tools change, we need to leverage
those changes. As processes change, we need to adapt. Even though we argue
about this sort of thing, modern editors help to keep the mistakes you
mention to a minimum.
Same goes for all the people who still shove str_ or int_ or p_ or
whatever in front of variable names. All that shows is that your coding
style is out of date. Now I have seen valid arguments where people preface
all variables with v_ regardless of type so that they sort together in
intellisense. I'm still trying to come to terms with that one :)
> I'm a barcode technologist and can tell you the typo error rate for
> numbers is at least 20 times lower than alphabet. It's not by accident!
By that logic we should all code in binary. Only two symbols :)
Ilus
On Sun, 29 Aug 2010 19:30:15 -0500, jwayt <jwayt_...@iobar.com> wrote:
jwayt
utterly passe'. I love this quote:
"Since the dawn of the commercial Web, technology has eclipsed content.
The new business model is to try to let the content ? the product, as it
were ? eclipse the technology. Jobs and Zuckerberg are trying to do this
like old-media moguls, fine-tuning all aspects of their product,
providing a more designed, directed, and polished experience. "
==========================================
WPF is about polished experience. It is however empty without real
content. The WPF aversion is due to its novelty as a medium. We
application painters (developers) have achieved some skill with our PB
(paint brush) and now we are going to need to work with a new pallet and
canvass. We all know (or should) that good content delivered by a good,
rich medium makes a good application.
Changing to the new medium will take time and effort to get over the
learning curve. I think Sybase has made that transition as easy as
possible. When trying to fit our square peg into the round hole, we just
may have to sand down its corners some.
Gov't distribution models diverge from commercial models that are
driving the technology, because they're goals are different. Gov't wants
wide distribution thru public channels/media and Commerce wants
private/tolled media stores. They do have a common concern: control of
delivery. Smart client can manage that delivery, so promises .Net.
The newsreader I'm using just told me there was an upgrade if I wanted
to get it. I had to download and install it. Yes, there was a DL
website, but after that, the client came up and I'm back in business.
That's a very workable model.
In article <8fsa76l6dq29meitf...@4ax.com>,
n0Spam...@sybase.com says...
>
> Haven't you guys heard? The web is dead!
>
> http://www.wired.com/magazine/2010/08/ff_webrip/all/1
>
> :-)
>
jwayt
MY POST WOULD HAVE BEEN PERFECTLY LEGIBLE IN UPPERCASE, BUT STYLE WOULD
SUGGEST I'M SHOUTING. i'm not.
You have described an editor that has eased the burden of retrieving
labels by targeting a reduced set of (appropriate) presented choices.
Once selected, the label comes into use in proper case. If Intellisense
can get case-independent cues, so much the better. If it reduces typing
and spelling mistakes, I'm for it.
In article <op.vifaj...@bhm-is-td-l1.corporate.local>,
Il...@gmail.com says...