dbTalk Databases Forums  

Bad bug in MVSP Insert()...

comp.databases.pick comp.databases.pick


Discuss Bad bug in MVSP Insert()... in the comp.databases.pick forum.



Reply
 
Thread Tools Display Modes
  #11  
Old   
Gene Buckle
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-03-2011 , 11:08 AM






To: Tony Gravagno
Tony wrote:
Quote:
From Newsgroup: comp.databases.pick

Gene, we're in sync. Preface all my responses here with "I hear ya
bud".

"Gene Buckle" wrote:
The -1 issue is embarassing. It NEVER should have made it out of beta with
such a fundamental flaw. I suspect the guy that wrote the code has either
never written code in Pick BASIC or was very, very new to it.

Or the code was written, tested, functional, and some subsequent
update broke things. I won't make excuses, and have no idea how
Take a look at the InsertCore() routine with ILSpy. Trust me, it's an
embarassing mistake.

Quote:
YMMV but I've been so turned off from contributing FOSS to this
community that it's really depressing.

It's the whole per-seat pricing mentality that makes it tough Tony. That and
a whole lot of "what's in it for ME?!" attitude. Frankly, I don't care either
way. If I deem it a worthy example of my work (and I get permission from my
employer), I'll post it. If someone benefits from it, great. If not, no big
deal.

By "per-seat" I'm guessing you're referring to mv.NET. I certainly
wasn't. Sure, I use and sell mv.NET but there are times when we don't
want or need it. Right tools for the job and all that. No, I was
seriously talking about FOSS and how projects like that notoriously
fail in this market.

Actually, I wasn't referring to mv.Net. I was referring to <insert per-seat
priced MV database product here>. mv.Net is a great product - it's just
not up to the kind of abuse I can put it to, especially when it requires a
latency adding middle-man server process.

Quote:
(Just making conversation, no real point...) As usual it comes down
to everyone taking, no one giving, and if you're lucky the "demand"
comes in the form of a few people just telling you what's wrong with
code rather than offering to send a cup of coffee for enhancements. I
have no problem with FOSS in general and I have a bunch of them in the
air - just trying to give back to the bazaar what I've consumed, so to
speak. Combining all this, I'm just saying the effort to publish a
wrapper lib for the D3CL as FOSS is a Lot of effort for a handful of
people who probably won't care, and with Very little reward except for
warm fuzzies. But it can never hurt to publish something you do and
hope that someone helps you to make it better just by pointing out
flaws that you might fall into with your own code.

Well the effort is going to be made one way or the other. I can no more stop
writing software than I could stop breathing. I can't NOT write software.
It's just the way I'm wired. If someone benefits from the code I produce,
great. If not, I really don't care. It's going to be written either way.

Quote:
per-seat priced solutions.

Me too. I think MVSP is kinda cool but I didn't want to get into it
TOO far until I saw others using it.

TL hasn't exactly trumpeted from the rooftops that MVSP even exists. Combine
that with a theoretically small number of shops that are both upgrading to v9
AND working with .Net applications and your potential audience ain't that
big. I need to shove harder to get them to create a version that is compatible
with Silverlight though - I could really leverage that to good advantage on
my production floor processes.

Quote:
Coming back to mv.NET though, MVSP (.NET/Java) is now in line with
UO.NET/UOJava, QMClient, and others. mv.NET uses those tools for
basic connectivity while providing a superset of higher functionality.
mv.NET (OK, take this as an [ad]) provides session pooling, code
generation, Silverlight integration, and an ADO.NET interface. These
I don't get the whole ADO.Net thing - that's designed to work with SQL type
data and (at least from where I sit) is fairly useless with MV datasets.
I will admit that using DataTables are a better way to add data to a grid
than the old AddItem() method of days gone by.

g.

--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.

ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!

Political correctness is a doctrine, fostered by a delusional, illogical
minority, and rabidly promoted by an unscrupulous mainstream media, which
holds forth the proposition that it is entirely possible to pick up a turd
by the clean end.
--- Synchronet 3.15a-Win32 NewsLink 1.91
The Retro Archive - telnet://bbs.retroarchive.org

Reply With Quote
  #12  
Old   
Tony Gravagno
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-03-2011 , 05:54 PM






"Gene Buckle" wrote:

Quote:
Take a look at the InsertCore() routine with ILSpy. Trust me, it's an
embarassing mistake.
I trust. I've seen and reported a couple other bugs that are similar
to the one you found.


Quote:
TL hasn't exactly trumpeted from the rooftops that MVSP even exists. Combine
that with a theoretically small number of shops that are both upgrading to v9
AND working with .Net applications and your potential audience ain't that
big.
Ultimately I think you'll come to see it my way on this...

MVSP is a basic connectivity library. That's how they're advertising
it. It's not supposed to be anything revolutionary, just the modern
evolutionary replacement for D3CL. As I see it, wasn't intended to be
much more than the obligator vendor-provided interface for that small
number of shops you mention. What's Very cool about it is that there
is both a .NET and a Java library, which responds to a lot of those
Linux-centric questions. But aside from that it's mostly just some
braindead but entirely adequate static classes.

I'm glad MVSP now exists. It's a good starting point for people who
want to learn .NET or Java, for someone who just wants free and simple
connectivity, or for people like us who might want to provide a free
or low-cost utility without a $260 surcharge for a connectivity
license. But for anything beyond the basics we just need something
more. That's where mv.NET picks up. I'm sure that sounds like a
sales pitch to people here, but as just another guy in the crowd, I'm
just telling ya that this is where my decision process takes me. YMMV


Quote:
I need to shove harder to get them to create a version that is compatible
with Silverlight though - I could really leverage that to good advantage on
my production floor processes.
I'll bet ya a Starbuck's mocha that you're not going to see
Silverlight integration from them. You said yourself the market is
almost too small just for uptake of the basic library. Not only is
Silverlight connectivity much more sophisticated, requiring much more
client/server development to make that work transparently, but the
number of people who would want to use it so far is tiny. Heck, TL
hasn't even implemented a basic mvItem class. Everything is string or
object. They have a long way to go just on the basic lib before
moving on to Silverlight - though I don't think this is the direction
in which they want to go.

Again, not trying to sound like a surrogate here, but mv.NET has had
Silverlight built-in for a while, and from my perspective only serious
larger shops are even interested. Silverlight integration is free in
mv.NET. Communications underpinnings, a XAML Wizard, and the business
objects code generator are all in there. The price of the software
hasn't changed in years and anyone who upgrades gets all of this at no
extra code. Even as a freebie I don't see uptake from the traditional
developer base.

What I'm saying is that the effort to implement these features is
huge, there's already a provider that does it competently at a low
cost, and TL by themselves doesn't have the market to justify
competing with a harmless third-party provider. They just want to
keep the core base happy and check the checkboxes. Yeah, they do .NET
and Java. Next!


Quote:
I don't get the whole ADO.Net thing - that's designed to work with SQL type
data and (at least from where I sit) is fairly useless with MV datasets.
I will admit that using DataTables are a better way to add data to a grid
than the old AddItem() method of days gone by.
mv.NET and UO.NET are intended for MV developers who understand
accounts, files, items, and callable subroutines. The ADO.NET libs in
both of these products are for .NET developers who don't know anything
about Pick, but need to see DataTables, Records, Columns, and Stored
Procedures. I have had clients who hire a .NET guy because it's too
tough to hire a Pick guy who also does .NET. Rather than training
them on Pick they just expose the DBMS in a way that makes sense to
the .NET guy. So now they can do databinding and exchanges with
relational systems without "foreign looking" code. I don't know a
single Pick guy, including myself, who uses the ADO.NET hooks on a
regular basis to get into the MV environment. But if someone asked me
to write an interface, and provide source, for some non-Pickie who
would later maintain my code - I'd use the ADO.NET mechanisms.
Similarly, if I needed to provide an encapsulated API to a non-Pickie,
and they needed to see a DataTable, it's dirt-simple with mv.NET to
say:
return myMvItemList.ToDataTable();
Poof, no loops to convert items to tables. (That said, I've created
my own extension libs to create different kinds of tables, datasets,
etc, but you get the idea...)

Anyway, different strokes for different folks. What I enjoy is that
these tools exist for us to discuss them - which is far better than
the alternative of no tools and/or no one to have the discussion.

Regards,
T

Reply With Quote
  #13  
Old   
Gene Buckle
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-04-2011 , 10:14 AM



To: Tony Gravagno
Tony wrote:
Quote:

I'm glad MVSP now exists. It's a good starting point for people who
want to learn .NET or Java, for someone who just wants free and simple
connectivity, or for people like us who might want to provide a free
or low-cost utility without a $260 surcharge for a connectivity
license. But for anything beyond the basics we just need something
more. That's where mv.NET picks up. I'm sure that sounds like a
sales pitch to people here, but as just another guy in the crowd, I'm
just telling ya that this is where my decision process takes me. YMMV


I've only got two issues with mv.Net. The first is that they charge a per-seat
fee for me to access data that I'm already paying someone else a per-seat fee
to access. Charging a per-seat fee for what boils down to a library is as
absurd as trying to price a compiler based on the number of programs they plan
on compiling with it. The second issue I have is the "server" in the middle.
All it does is add more network latency. I don't need connection pooling. I
don't _want_ connection pooling. What I need is eye-blistering speed.

They should take a look at how Infragistics, Leadtools and LMD do things.

Quote:
I need to shove harder to get them to create a version that is compatible
with Silverlight though - I could really leverage that to good advantage on
my production floor processes.

I'll bet ya a Starbuck's mocha that you're not going to see
Silverlight integration from them. You said yourself the market is
almost too small just for uptake of the basic library. Not only is
Silverlight connectivity much more sophisticated, requiring much more
client/server development to make that work transparently, but the
number of people who would want to use it so far is tiny. Heck, TL
hasn't even implemented a basic mvItem class. Everything is string or
object. They have a long way to go just on the basic lib before
moving on to Silverlight - though I don't think this is the direction
in which they want to go.

You're probably correct. I'll likely end up just doing it myself if
I feel the need is there. I don't expect it would be too difficult to shoe
horn the existing MVSP code into a Silverlight compatible library.

Quote:
Again, not trying to sound like a surrogate here, but mv.NET has had
Silverlight built-in for a while, and from my perspective only serious
larger shops are even interested. Silverlight integration is free in
mv.NET. Communications underpinnings, a XAML Wizard, and the business
objects code generator are all in there. The price of the software
hasn't changed in years and anyone who upgrades gets all of this at no
extra code. Even as a freebie I don't see uptake from the traditional
developer base.

That surprises me. Silverlight makes for a great presentation layer and being
able to offload all the heavy lifting on to the server via rule modules is a
huge win in my book. Then again, I look at things through the filter of being
an ERP geek.

Quote:
both of these products are for .NET developers who don't know anything
about Pick, but need to see DataTables, Records, Columns, and Stored
Procedures. I have had clients who hire a .NET guy because it's too
tough to hire a Pick guy who also does .NET. Rather than training
*waves hands* Right here! Pick & .Net wrapped up in the same geektastic
package. *laughs*

Quote:
them on Pick they just expose the DBMS in a way that makes sense to
the .NET guy. So now they can do databinding and exchanges with
relational systems without "foreign looking" code. I don't know a
single Pick guy, including myself, who uses the ADO.NET hooks on a
regular basis to get into the MV environment. But if someone asked me
to write an interface, and provide source, for some non-Pickie who
would later maintain my code - I'd use the ADO.NET mechanisms.
Ahh, that makes sense. Now me, I'd be pushing the ADO.Net types to learn
Pick. If nothing else, it would help reduce the number of people dependant on
SQL. *huge grin*

Quote:
Anyway, different strokes for different folks. What I enjoy is that
these tools exist for us to discuss them - which is far better than
the alternative of no tools and/or no one to have the discussion.

Agreed. I'm debating on how nuts it would be to build a version of MVSP in
Delphi XE. THAT would scream. You can't beat a native compiler.

g.

--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.

ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!

Political correctness is a doctrine, fostered by a delusional, illogical
minority, and rabidly promoted by an unscrupulous mainstream media, which
holds forth the proposition that it is entirely possible to pick up a turd
by the clean end.
--- Synchronet 3.15a-Win32 NewsLink 1.91
The Retro Archive - telnet://bbs.retroarchive.org

Reply With Quote
  #14  
Old   
Gene Buckle
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-05-2011 , 11:39 AM



To: Tony Gravagno
Tony wrote:
Quote:
From Newsgroup: comp.databases.pick

"Gene Buckle" wrote:

I've only got two issues with mv.Net. The first is that they charge a per-seat
fee for me to access data that I'm already paying someone else a per-seat fee
to access.

But "per-seat" with mv.NET is Way different from per-seat DBMS
But it's _still_ per-seat. If I was doing Internet facing applications that
required session pooling in order to maximize the seats I've got at the D3
end, I'd certainly look at mv.Net again.

Quote:
But you're the only person I know who says that kind of performance is
more important to you than overall cost and all of the other features
I've mentioned.

Because of the environment I'm working in, those "other features" really
don't do much towards getting the tasks I need done, done. Sure, mvItem
would be nice to have, but the overhead in both price & latency isn't
worth the hassle. If I _really_ have to have what mvItem does, I can write
that myself. I write software that supports manufacturing, accounting and
logistic processes. My user base is small (~250 users) and not only
technically competent, but pretty demanding as well. They deserve the
absolute best I can give them and building fast systems is part of that.

Quote:
Further, you say you don't want pooling and you do need speed. But in
many scenarios, not all, those factors counter one another. Pooling
is used to improve performance for most new development scenarios that
people consider these days. It's basic parallel processing vs serial.

In _my_ situation, pooling is just a mother-may-i layer that adds nothing
but network latency. Are the developers you're talking to targeting
the browser or are they writing thick and "medium" clients?

Quote:
You have to remember that this MV market is so small that all vendors
need to consider unpopular license models just to stay in business.
I lay this one right at the feet of the MV vendors themselves. The tool
vendors wouldn't have to use "unpopular" models if they were doing a hell
of a lot more to promote the uptake of MV in general. They need to do
more than go "MultiValue is better than SQL!" They need to shove that fact
into peoples' faces all the time and never let up. Otherwise, they're going
to be dead in another decade. I think the only one that has a chance is
Ladybridge and that's because they're in a great position to help those that
are left high & dry by the death of a primary vendor and are in dire need of
a best-of-breed replacement.

Quote:
I don't expect it would be too difficult to shoe
horn the existing MVSP code into a Silverlight compatible library.

If you undertake that, please email me. I'm curious how you'd
approach it. Essentially I think you'd need a faade/proxy class on
the client which uses a web service to connect to a remote server.
The server then does the actual processing. What's neat about this is
that you might be able to get your data from a *nix server while the
UI is served via ASP.NET from Windows. I don't know if that's a good
idea in general but offloading the data access to another server,
especially Linux, could have some performance benefits. Just a
thought.

The shortest path would be a Silverlight based implementation of the MVSP wire
protocol itself. It would work well in an Intranet-only situation which is
what I'd be working with.

Quote:
huge win in my book. Then again, I look at things through the filter of being
an ERP geek.

(BTW, that was supposed to be "no extra COST", not "no extra code".)

Agreed. I think part of the problem is that Silverlight has undergone
a slow evolution from v1 to v4 and only now are real gear-heads using
it for RIA, line-of-business apps like we have in this market. In the
MV world people equate Silverlight to Flash, think glitz, and turn the
other way. Other people have gotten fed up with too many options,
from ASP, to ASP.NET web forms, to ASP.NET MVC, and now Silverlight
and it's radical client/server model. Add to that all the hullabaloo
about plugins vs HTML5 and traditional Pick sentiments, and you get a
hugely confused audience that is very reluctant to adopt anything
anymore.

I tend to look at Sliverlight as if it were basically a browser-hosted version
of WinForms. The NetAdvantage Silverlight controls makes it pretty painless.

Quote:
I'm debating on how nuts it would be to build a version of MVSP in
Delphi XE. THAT would scream. You can't beat a native compiler.

I really need to publish my work on mvEsperanto. I think you would be
VERY interested.

Hehe. I bet it sounds better in the original mvKlingon. *laughs*

Quote:
Thanks for the discussion bud. I'm enjoying this one. (And wondering
just how many people will actually read this far. HAHA)

We probably should have offered a prize if they get this far.

g.

--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.

ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!

Political correctness is a doctrine, fostered by a delusional, illogical
minority, and rabidly promoted by an unscrupulous mainstream media, which
holds forth the proposition that it is entirely possible to pick up a turd
by the clean end.
--- Synchronet 3.15a-Win32 NewsLink 1.91
The Retro Archive - telnet://bbs.retroarchive.org

Reply With Quote
  #15  
Old   
Tony Gravagno
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-06-2011 , 12:57 PM



"Gene Buckle" wrote:

Quote:
Further, you say you don't want pooling and you do need speed. But in
many scenarios, not all, those factors counter one another. Pooling
is used to improve performance for most new development scenarios that
people consider these days. It's basic parallel processing vs serial.

In _my_ situation, pooling is just a mother-may-i layer that adds nothing
but network latency. Are the developers you're talking to targeting
the browser or are they writing thick and "medium" clients?
I haven't had a single request for a thick or even medium-thick client
for years. Everyone wants thin client. Of course that means chatty
AJAX which can degrade some performance so I'm compelled to make up
for it with some scripting. There's a constant dance in there to
ensure the environment is performant, maintainable, and cross-browser
capable.

Since everyone I talk to wants thin-client with a browser UI, we
rarely have any idea of how many "users" there will be. This throws
traditional per-seat counts out the window. So we try to keep every
postback or AJAX hit tiny so that we can process through them as fast
as possible. With a single DBMS license and corresponding mv.NET
license we can frequently get a realistic 5:1 throughput. As we add
more DBMS and mv.NET licenses the load gets spread out and we can get
closer to 20:1. So with a non-pooled environment like this, you might
need 20 DBMS licenses to process maybe 60 users. While I might only
need 3 DBMS licenses plus 3 mv.NET licenses for the same load. If you
Don't allocate those 20 licenses to handle the 60 users, you're going
to incur some serious network latency as more people back up in line
to process on fewer lines. It's the old grocery store checkout line
scenario - "aisle 3 is open, I can help you over here". These
numbers are admittedly contrived but I think the scaling factor is
fairly consistent. The bottom line is that most sites can save a lot
of money with pooling.

Again, for anyone concerned about DBMS vendors surviving this, the
bottom line is that they keep sites that can operate economically and
they lose sites that cannot. Everyone needs to work harder to produce
the same revenue we produced from DBMS licenses a decade ago but DBMS
vendors seem to think they're going to get the same sales counts using
the same licensing model they used in the 1980's.

Honestly I haven't seen a single site reduce their DBMS license count
after adding a GUI, though this is usually the goal from the start.
What we see is that sites add a few DBMS licenses to support external
GUI/BUI users while keeping most internal people on the CUI. As
described, the licensing costs for this are relatively painless, and
sites gladly purchase a few more licenses than to migrating to Oracle
or some other platform. After all of this rhetoric, that's really how
I see this play out in almost all situations. Gene, you're just a
rebel.




Quote:
You have to remember that this MV market is so small that all vendors
need to consider unpopular license models just to stay in business.
I lay this one right at the feet of the MV vendors themselves. The tool
vendors wouldn't have to use "unpopular" models if they were doing a hell
of a lot more to promote the uptake of MV in general. They need to do
more than go "MultiValue is better than SQL!" They need to shove that fact
into peoples' faces all the time and never let up. Otherwise, they're going
to be dead in another decade. I think the only one that has a chance is
Ladybridge and that's because they're in a great position to help those that
are left high & dry by the death of a primary vendor and are in dire need of
a best-of-breed replacement.
I wrote my above notes before I read your note here so you see we're
in agreement about vendor responsibility.

However, while I agree that QM is a great platform, I'd submit that
all of the MV providers are eager and capable of migrating competing
systems. Best-of-breed once again depends on requirements.
Personally, for small projects I like QM but for larger ones I'm
inclined toward the larger providers, and the selection of which
provider is dependent on my technical requirements.


Quote:
I don't expect it would be too difficult to shoe
horn the existing MVSP code into a Silverlight compatible library.

If you undertake that, please email me. I'm curious how you'd
approach it. Essentially I think you'd need a faade/proxy class on
(Looks like Google or some other system in the middle doesn't speak
french, that was "fac'ade/proxy class" with a non-english 'c'. )


Quote:
the client which uses a web service to connect to a remote server.
The server then does the actual processing. What's neat about this is
that you might be able to get your data from a *nix server while the
UI is served via ASP.NET from Windows. I don't know if that's a good
idea in general but offloading the data access to another server,
especially Linux, could have some performance benefits. Just a
thought.

The shortest path would be a Silverlight based implementation of the MVSP wire
protocol itself. It would work well in an Intranet-only situation which is
what I'd be working with.
Hmm, I'd check with TL to see if they've tried MVSP in Silverlight.
The modest requirements of that small library might not need more
than what's available in that environment.

If they say it won't work then you're talking about reverse-
engineering the wire protocol. You really don't want to go there.
Just create a proxy and wire it up yourself.

Since you're talking about creating a D3CL wrapper anyway, just do
that but rather than hardcoding MVSP into the implementation, abstract
out the connector piece so that you can use MVSP locally OR proxy over
to a server. I've already done something like that. I have a single
library that I use for connectivity with MVSP, mv.NET, QMClient, and
SSH without an underlying library. I code to my interface, not
theirs. In your case you'd code to D3CL, and you'd be able to swap
out the comms layer for whatever you want later, at any tier.


Quote:
I tend to look at Sliverlight as if it were basically a browser-hosted version
of WinForms.
In some cases it is but in others not. If you're maintaining state
with a rich set of classes you probably don't want all of that data in
the client, and from a DBMS access perspective you may not want to
expose your DBMS to a direct connection from a remote client anyway.
So many Silverlight solutions are client/server implementations where
the client proxy accepts requests which are transparently forwarded to
the server for processing.

I should say I am FAR from being a Silverlight expert so I'm nearing
my limits on implementation details here. From my perspective, from
v1 to v3 Silverlight has been a moving target and inadequate for true
RIA as we would expect for our apps. Only in v4 does it now seem
robust enough, and of course in the latest version there have been
paradigm changes to make all that possible. So I'm seriously behind
the curve, but still capable of keeping my end up to this point.


Quote:
The NetAdvantage Silverlight controls makes it pretty painless.
There ya go, an example of 'you get what you pay for'. MVSP is basic
connectivity for free. If you want a "pretty painless" experience
with something more sophisticated like Silverlight, you will either
need to make or buy it.

"You can have it good, cheap, or fast - pick two."

T

Reply With Quote
  #16  
Old   
Gene Buckle
 
Posts: n/a

Default Re: Bad bug in MVSP Insert()... - 10-07-2011 , 10:15 AM



To: Tony Gravagno
Tony wrote:
Quote:
From Newsgroup: comp.databases.pick

"Gene Buckle" wrote:

Further, you say you don't want pooling and you do need speed. But in
many scenarios, not all, those factors counter one another. Pooling
is used to improve performance for most new development scenarios that
people consider these days. It's basic parallel processing vs serial.

In _my_ situation, pooling is just a mother-may-i layer that adds nothing
but network latency. Are the developers you're talking to targeting
the browser or are they writing thick and "medium" clients?

I haven't had a single request for a thick or even medium-thick client
for years. Everyone wants thin client. Of course that means chatty
AJAX which can degrade some performance so I'm compelled to make up
for it with some scripting. There's a constant dance in there to
ensure the environment is performant, maintainable, and cross-browser
capable.

That's interesting. The only problem I've got with thin-client is that
I can't get the same interface functionality as I can with a thick-client.
By using the Click-Once deployment offered by .Net, it eliminates the update
problems normally associated with thick-clients. Silverlight 4 offers me
a very good "Winforms" like experience and that's why I'm looking further
into it.

Quote:
GUI/BUI users while keeping most internal people on the CUI. As
described, the licensing costs for this are relatively painless, and
sites gladly purchase a few more licenses than to migrating to Oracle
or some other platform. After all of this rhetoric, that's really how
I see this play out in almost all situations. Gene, you're just a
rebel.

Yep. I drive my AS/400-laden co-workers out of their minds sometimes.
Mostly from things like:

Them: We need to do <foo> by <date>.
Me: Ok.
Them: <discussions about how they're going to do their bit>
Them: Gene, will you be able to do <foo> by <date>?
Me: I did it while you were talking about how you were going to do it.
Them: Wha?
Me: *huge grin*

They get annoyed when I tease them about their cryptic table names and length-
limited field descriptions.

Quote:
However, while I agree that QM is a great platform, I'd submit that
all of the MV providers are eager and capable of migrating competing
systems. Best-of-breed once again depends on requirements.
Personally, for small projects I like QM but for larger ones I'm
inclined toward the larger providers, and the selection of which
provider is dependent on my technical requirements.

Do you thing QM is slower than the larger players (D3, Reality, etc)?

Quote:
I don't expect it would be too difficult to shoe
horn the existing MVSP code into a Silverlight compatible library.

If you undertake that, please email me. I'm curious how you'd
approach it. Essentially I think you'd need a faade/proxy class on

(Looks like Google or some other system in the middle doesn't speak
french, that was "fac'ade/proxy class" with a non-english 'c'. )
I read news with tin and it bitches about any DBCS characters. You're likely
writing your original in HTML so it includes the fancy characters. I have to
remove them in order to get tin to post the reply.

Quote:
The shortest path would be a Silverlight based implementation of the MVSP wire
protocol itself. It would work well in an Intranet-only situation which is
what I'd be working with.

Hmm, I'd check with TL to see if they've tried MVSP in Silverlight.
The modest requirements of that small library might not need more
than what's available in that environment.

If they say it won't work then you're talking about reverse-
engineering the wire protocol. You really don't want to go there.
Just create a proxy and wire it up yourself.

It won't be that hard. It might even be fun to build an MVSP server daemon
for QM.

Quote:
The NetAdvantage Silverlight controls makes it pretty painless.

There ya go, an example of 'you get what you pay for'. MVSP is basic
connectivity for free. If you want a "pretty painless" experience
with something more sophisticated like Silverlight, you will either
need to make or buy it.

I don't mind buying it - I mind buying it again and again and again.

g.


--
Proud owner of F-15C 80-0007
http://www.f15sim.com - The only one of its kind.
http://www.diy-cockpits.org/coll - Go Collimated or Go Home.
Some people collect things for a hobby. Geeks collect hobbies.

ScarletDME - The red hot Data Management Environment
A Multi-Value database for the masses, not the classes.
http://www.scarletdme.org - Get it _today_!

Political correctness is a doctrine, fostered by a delusional, illogical
minority, and rabidly promoted by an unscrupulous mainstream media, which
holds forth the proposition that it is entirely possible to pick up a turd
by the clean end.
--- Synchronet 3.15a-Win32 NewsLink 1.91
The Retro Archive - telnet://bbs.retroarchive.org

Reply With Quote
Reply




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off



Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.