www.justiceforchandra.com Forum Index www.justiceforchandra.com
Justice for Chandra Levy and missing women
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

rdwrites iseries blog
Goto page Previous  1, 2, 3, ... 10, 11, 12  Next
 
Post new topic   Reply to topic    www.justiceforchandra.com Forum Index -> www.rdwrites.com forum (part of www.justiceforchandra.com)
View previous topic :: View next topic  
Author Message
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Mon Apr 12, 2010 7:35 pm    Post subject: Reply with quote

from systeminetwork.com:
This activity continues. I have found three messages in my inbox that looks like spam, the last posted yesterday by "RobertTaylor".

Spam and phishing type post. Those links can be dangerous and loaded with viruses.

Yes, very dangerous to click on the link they demand you click on or else they will "report to you to your ISP". These are criminals of the worst kind, carjacking and worse types, they have only one intent - to turn your computer into a zombie and steal everything worth stealing they can get from it and everything you type in, including logins into work or anything with value such as military/industrial spying.

I got one too finally, from so called RobertTaylor. Based on what has been coming into my site, this is at the level of warfare. That is not hyperbole. The range of addresses used in the attacks is unprecedented, but mostly from Russia and China and everything around them. The number of PC's in the US they have zombied is huge also.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Tue Apr 13, 2010 11:15 pm    Post subject: Reply with quote

I wouldn't want to be likened to Roger Covey of SSA. I never knew him personally, but it seems that press has turned negative, against him.

There were quite a few S/3x vendors that grew very quickly in the 80's and 90's. They were riding a wave as small to medium businesses were migrating en mass away from time-share providers like EDS, or purchasing vendor supplied solutions to replace dysfunctional in-house systems.

BPCS may have begun as an incubator project, starting out on a kitchen table, but it didn't take much in the early 80's to be able to offer something better than so many cobbled together, poorly designed in-house systems.

Today is a different story. How would an incubator project be able to compete against so many mature, robust, broadly scoped, heavily modified enterprise class systems? Supported by heavily armed (very large) solution providers?

I actually do have some ideas about how incubator projects may be able to compete against more mature offerings, but it's not easy. And if you're hoping for a modern repeat of the type of growth that SSA experienced, then you might be sorely disappointed.

-Nathan


I only saw Covey once in the development area in the time I was on the 17th floor at SSA (I spent another year on off-site development for them). He was posting QUIET signs. Never heard of anyone getting a warm and fuzzy feeling from him, but he did build that company from his kitchen table.

Easier then than now? I'm sure it wasn't so easy then either. What made the SYS/3X software work though was high performance business software and interface that scaled relative to what else was available.

No matter what one starts with on the kitchen table, that's the most important thing.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Fri Apr 16, 2010 10:29 pm    Post subject: Reply with quote

I was going to skip reading the RPG Open I/O doc but ended up looking it over. The OA handler handles both 5250 and anything else intended to support such as browser. But the OA chargeble portion is embarrassingly trivial. This as some kind of amazing data buffer API product is the dumbest thing I've heard lately.

Having said that, I'm going to list all the nice things I can about it.

The RPG aspect of it and the doc is simple, elegant, and excellent, all trademarks of Barbara Morris. I do things with RPG nearly everyday that I haven't done before, after 21 years of RPG programming starting with RPG/400, and each time I do something I am yet again amazed at the power of the ILE RPG compiler. There's only one person at IBM whose work I know about and respect, and that's Barbara Morris.

Concerning the OA architecture, requests to support other interfaces with RPG I/O have been made from the beginning of seeing the need to be able to support client/server and then browser interfaces, and to be able to do so leveraging existing skills. The OA I/O exit call isolates interface programming automatically with file or screen I/O commands which can be multi-purposed in ways unimaginable at the time of coding a CHAIN or UPDATE or EXFMT in an RPG program.

And lastly, IBM has a vision for RPG.

Now for what I don't like about it.

IBM has a vision for RPG.

Between 5250 interactive charges, CPU governors, trying to force their customer base to Websphere and Java with onerous charges for frozen and continually threatened discontinued existing technology, making license transfers illegal, attempting to force people to prop up their Rational software, and in general squeezing customers until they leave because they can, this latest vision from IBM beancounters of making a trivial API a product with a runtime license no less scares the bejezus out of me.

If that is IBM's vision for the future, the future is very bleak indeed.

I could go on about how you can call a program instead of the handler exit call calling it for you, but I'm afraid they'll make it a chargeable feature with a runtime license, otherwise formerly known as licensing OS/400 or i5/OS, so I'll stop lest I teach the beancounters an idea for a new runtime license

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Fri Apr 16, 2010 10:45 pm    Post subject: Reply with quote

from systeminetwork, Henrik Rutzou wrote:

Constantly rebranding gives you no brand at all.

well said, and indeed the only brand they have is from the past for what we use and love. I'm sure that includes the mainframe people as well and I doubt AIX people do anything but scoff at p.

They have no brand at all. And indeed, that seems to be the intent.

As for what I and other people call it, I refused to be poked and prodded into their silly word games, and now letter games, and referred to it in posting as AS/400 far after it became unfashionable to do so. That's what many of our users call it. That's what they know it as. That's what it is. Quite frankly a genius architecture IBM won't even talk about.

But finally after several iterations of i themed names I settled on iseries in the self referencing sense as in PHP, a series of i names including iseries itself.

So iseries works for me for as long as they have stupid marketing people at IBM.

iseries is forever.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Sat Apr 17, 2010 7:54 pm    Post subject: Reply with quote

from systeminetwork,

maybe a handler and RPGOA to support XWindows?

The handler is just an ILE program written by someone, anyone. It has no magical powers. It's just an exit call program, with the exit being an I/O command on a F spec in RPG with the HANDLER keyword specifying the exit program to call.

XWindows is a very high overhead protocol but I have ranted for years that only the AS/400 is commonly said not to have a GUI while Unix is presumed to have the XWindows GUI, even though we have had it as well. The bottom line is there is no difference between Unix and OS/400 in capability, AIX is a subset of OS/400 and OS/400 is POSIX compliant and has been for decades, yet Unix somehow gets this ride and OS/400 is called legacy.

It's a situation partly IBM's making. They push AIX and develop products for it that they say is not available for OS/400 and in general act like they want to shut down Rochester because they had the temerity to produce a best selling product on their own.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Mon Apr 19, 2010 7:48 am    Post subject: Reply with quote

IBM starts off the week irritating me. TMP at itjungle quotes Ian Jarman as saying he expects handlers will be written in C/C++.

Typical elitist bs. If we wanted/needed to write something in C/C++, we would. We don't need a runtime exit call to let IBM and real programmers do hard things for us that we just can't understand, blah blah blah.

C/C++ is just another language in our ILE arsenal, one I haven't needed yet. The IBM elitist implication is that real programmers who write handlers will write in a real programming language we already have, and by implication that RPG isn't good enough.

Irritating.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Mon Apr 19, 2010 1:06 pm    Post subject: Reply with quote

Henrik wrote:
why should handlers be written in C, just because they are named "handlers". Most of these handler will talk to devices over HTTP and that means that they just has to "repack" and add metadata (wherever they may come from, the sky ?) to a DS to HTML, XML, JSON or something similar.

Chris wrote:
I do not intend to get involved in a why is language A better than language B, as they says its horses for courses! The documentation I have seen says you can use ANY ILE language and IBM has supplied the include files in QOAR for you to do so.

well, that's the answer, Chris, you're right. Had IBM said that that would have been fine.

But they didn't say it because they're trying to justify a runtime license by implying that this is a gateway to a low level driver handling stuff we can't do in RPG. In other words, typical IBM iseries marketing malfunction.

I would use C in a heartbeat for something where it provided a compelling advantage, but unlike Java RPG has pointer manipulation as in C and RPG handles packed and integer equally well. However C modules can be bound in as desired, I just think there has to be a compelling advantage to mix in another language to an RPG program.

Also the handler has to handle 5250 as well as any other communications such as socket driven protocols and CGI/web interfaces. The whole architecure doesn't make a lot of sense to me compared to more open ended dataq processing but I'll not criticize IBM for that, they have been repeatedly asked to provide seamless alternative communications to 5250 programs and this does that.

I think isolating out business logic to dataq server programs from various interface programs such as CGI, 5250, and sockets makes a lot more sense. I have been working on that for awhile and finally getting up a head of steam on it.

This initiative could have positioned the iseries as a multi-interface system instead of a system with allegedly no GUI, but the hamhandedness of locking the customer into buying a runtime for third party or self written interfaces did more damage than the initiative did good. For anyone that really wants to do this I don't see a huge benefit for OA over replacing an EXFMT or file IO with service program subprocedures, ala the handler.

But I think dataq server programs are the way to go for interface independent programming anyway.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Tue Apr 27, 2010 3:34 am    Post subject: Reply with quote

Here's the Pros for RPG Open Access:

Buying latest version of IBM ILE development software -
allows customer with source code to their programs to disable any file access (display files and/or physical files and/or printer files) in an RPG program and hardcode the name of a program to call instead of performing the IO command.

Buying a product from a vendor -
allows you to have a program name to hardcode into the RPG code on the line of the file you want to disable to let the product perform alternative actions.

Buying a runtime license from IBM -
allows IBM to allow the disabled file to be passed to a product you paid big bucks for to do whatever you paid them for.

With pros like that, who needs cons?

I reject the concept that an RPG shop can't go through their code and comment out an EXFMT and call a program, vendor product or open source, passing data structure of the file and statuses. It's not that RPG shops can't or don't want to make a program call, they don't have a program to call.

Having said that, I want every one of these vendors and consultants to succeed with their products. How many freakin GUI products do we have to have before people stop saying we don't have a GUI interface? I think we'd put about any other server system in existence to shame with our plethora of GUI products, every imaginable niche, tier, and yes now even run time, it's all covered, from screen scraping to e-server CGI to CGIDEV2 to PHP to Java JSP to Java J2EE implemented in every imaginable possibility from open source to CASE development platforms.

And I hope they all do well, including this latest tweak from IBM. It's got its pros, just like every other possible solution. By the time an iseries shop buys into one of these vendor solutions, the run time license is peanuts, I'm sure. IBM essentially has a runtime license on other solutions such as Websphere, so as always we have to add everything up and see if its a compelling value.

OA adds a pricey tweak to our already innumerable GUI options. I hope it wins over and keeps top tier customers. But I think it will take viral open source iseries solutions to appreciably increase our customer base again.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Fri Apr 30, 2010 3:52 am    Post subject: Reply with quote

I read the new IBM white paper on iseries strategy. This is an excellent executive summary white paper on the value of the system i (I think that's what it is called in there. Very forgettable. They refer to AS/400 and iseries enough to make it understandable though.

The bias for system p, oh excuse me, they refer to it as AIX for some reason, didn't they get the memo?, is not overwhelming as usual. They actually had some people who know something about AS/400 land working on this. Congratulations.

Also as usual, they make the AS/400 iseries system i IBM i (and we won't even go into the OS names, oh wait, does the OS even have a name now?) come off as simplistic with less expensive to operate based on less IT personnel and data center costs. IBM is averse (nay, make that would never) refer to any operating system architectural advantages for fear of upsetting their Unix and mainframe predominating culture. IBM i is simple. Yeah, that's inoffensive enough.

They do talk very well about the enterprise level class of iseries customer. Also explain well the makeup of customer base, sales, and likely iseries environments (everything but weather forecasting and oil geographic data analysis pretty much sums it up).

IBM give excellent summaries of new features in 6.1 and 7.1, and a very good contrast in iseries hacking and virus security versus Windows, while being 41% less expensive to run your business on.

This is a good start.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Sat May 01, 2010 7:19 pm    Post subject: Reply with quote

I don't want to begrudge IBM making money. The i operating system is a quality OS, the best there is I think any iseries person will agree, and IBM keeps adding quality to it with every release.

RPG Open Access could play out as an extremely marginal pricey tweak for a small subset of iseries customers with source code to their 5250 interactive programs but uncomfortable commenting out an EXFMT and replacing it with a call to a third party program. The number must be exceedingly small.

The point of webfacing was that recompiling the interactive programs wasn't an attractive option, for reasons that could vary from not having known to be reliable source code to regulatory reasons of having to retest and validate for production every interactive program the business has.

That the iseries customer has the source code and ability to recompile but doesn't have the ability to comment out EXFMT commands and replace with a call passing the screen data structure may be some kind of belief lurking in decision making places, but is not true. Once you have an example of one call, they are all the same.

Further to that fuzziness in decision making places is the typical bias toward Unix and a belief that C/C++ is a real language, epitomized with the IBM statement that RPG Open Access handlers will typically be written in C/C++. That rises as much passion in me as the insulting API charge.

A minimalist appraisal of this product is that IBM is charging a premium for an API that allows web facing to not have to figure out what's on the screen. Once figured out for each screen, anything can be done with data ala the handler as having the data passed to you in a data structure.

My minimalist appraisal is that IBM figures that if they add an API to the i operating system (the OS/400 - i5/OS operating system where API's were added as the Supreme Being intended) that supplants need for the webfacing product, then they will replace that revenue with an API charge. That is just embarrassing.

Imagine if IBM tried this with Linux. Oh wait, they can't, Linux is open source. Imagine if Microsoft charged for driver API's, which is what this sort of is but even less than that. Well, they do charge for API development tools but imagine if every Windows customer, or even just the businesses that are deserting IBM in droves to replace the iseries with Windows, had to pay a "run-time" license to Microsoft to use a Windows program that used one of Microsoft's precious API's. You can imagine how that would go over. On the other hand, Microsoft would probably create a server for it and you'd have to buy another Windows server instead. So I'l let that analogy rest there.

But this could play out entirely differently, and I truly hope it does. The thing that has hurt us the most has been lack of a standard GUI interface, and this piggybacks the standard interface exceedingly well. Albeit it is within 5250 data space, and that very few web pages will match up with an intense 5250 screen as our users want, and that in reality the interactive programs will have to be rewritten to match up more with the various web page tabs and AJAX calls that can come in, this is a standard interface that makes third party interfaces transparent. That's an exceedingly good thing.

I will add that the actual 5250 external screen DDS equivalent was done correctly with CGIDEV2, but although CGIDEV2, PHP, and other open source solutions such as Java servers are still the probable technologies that will drive iseries web serving use, it would be great if something like Profound UI and equivalents from others are seen as a standard interface for the iseries even though different third party products are involved, because programming is standard, not to mention the same powerful programming that made the AS/400 what it is.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Tue May 04, 2010 2:31 pm    Post subject: Reply with quote

I have to amend my post above. Just got out of the IBM Q&A session at COMMON where I asked about the OS names. Answer was IBM i and AIX (and Linux) are current names, there is no system i and system p anymore, they say.

I was quite surprised. The whole letter thing i, p, x, and z went away? But when it did, p got its natural name AIX back, but we're stuck with i.

Maybe the Unix people just told them to shove it and took their name back?

IBM seems quite pleased with IBM i as a name, pointing it out on their badges and literature as one of the three OS'es on the Power (something goes here, System? who knows).

I had more important things to ask about their lack of addressing capabilities and architetcure of IBM i, and I did, but first I had to ask what it was I was asking about.

I think they gave a genuinely heartfelt response to the internal recognition of the capabilities of the OS/400 genesis operating system and will keep it in mind, and that's all I can ask of a Q&A session that IBM clearly showed their commitment to in who they sent.

So where I post above about AIX still being called AIX for some reason, the memo has come and gone and been replaced with more memos, and it's AIX yet again.

Should we be so lucky to get our recognizable name back again like the Unix people pulled off.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Wed May 05, 2010 1:10 pm    Post subject: Reply with quote

I just got out of the ProfoundUI session here at COMMON 2010 in Orlando, and listened to Alex Roytman give his presentation. I also looked at Nathan's post (link is working now) and Nathan is right about what ProfoundUI does in building a DDS.

I disagree with Nathan profoundly, however, on binding of web page data to database fields as being a negative. It is exactly the same as our DDS reference keyword, and I use that and the *LIKE reference in RPG programs for exactly that, a change in data type / length automatically picked up during the compile as a key to maintainable code.

The elegance and integration of the ProfoundUI with RPG / DDS programming is as if all the brilliant IBM'ers involved in architecting OS/400 and RPG DDS and then ILE extended the architecture to add a standard GUI interface to the AS/400. This is the way it should have been done had they attempted it. This is as standard a GUI interface to the iseries as could ever be done.

The immediate reaction will be, well, IBM doesn't own it, it doesn't ship with the IBM i OS, so it can't be standard. Plus IBM half the time really wants you to use cross platform Websphere Java server, and in that sense wants Websphere to be the IBM "standard" across all their servers. And there are those who have chosen other 4GL approaches closely integrated with the iseries such as BCD who find more productivity and flexibility in that approach who may not want the costs of a standard GUI embedded in the cost of the OS. That's fair enough.

I would add ProfoundUI to an elite group of products that every iseries should have, Robot and DBU. It's in the must have standard category is the way I would put it.

Having said that, ProfoundUI fits the traditional iseries RPG program serving role of complex business programming filled by 5250. Given that it has a job / program per user as in 5250 QINTER but instead in QHTTPSVR, it's not typical for the high volume / low content of traditional web programming such as content sites. And that hasn't been our typical customer base in any event.

I would also add that my personal interests still lie in CGIDEV2 and an open interface, I will be looking at Aaron's open UI project, but I think that any iseries shop that does 5250 programming should factor in ProfoundUI licensing and IBM's RPG Open Access licensing as part of the cost of the iseries and use it. It is that fundamental to the core of what we do.

Concerning RPG Open Access, an equivalent could be done of data structures passed to programs or service program subprocedures instead of EXFMT and RPG Open Access. If one is so inclined though I believe they would have already done something along those lines.

There is also at least one CGIDEV2 based open source rich web interface that can be used instead of paying any additional licensing. But with ProfoundUI, there is no HTML, CSS, or Javascript coding. Development is with a Delphi or VB type screen designer with properties, etc. as all IDE's have these days.

However, the ProfoundUI Screen Designer is totally i/OS RPG / DDS aware, runs from the iseries as wuth rdi, and properties are sensible for subfile type programming and other RPG DDS architectural features. An RPG programmer will be laying out and creating web pages displaying data from their RPG programs and their DB2 files almost instantly instead of setting up WRKSTN DDS for 5250 screens. Then they write RPG programs exactly as they do today and put a HANDLER keyword on the WRKSTN F spec that ProfoundUI creates from their layout, and their screen IO is then to web pages.

The web pages you build in the designer can have as many widgets added to them as you want, making it as rich and interactive as anything created anywhere. The product converts existing DDS for you as a base and will webface unconverted 5250 screens in the interim. During the session many questions were asked, and every question of can we do this was answered yes with an explanation and in some cases a demo.

Speaking of demos, the demo was done in Orlando against an iseries in Ohio, and screens were immediate, essentially instantaneous. This includes keystroke level interactivty used in Web 2.0 RIA feutures of the design, such as selections when keying in something. Subfile type data could be loadall or page by page. I asked if this scales as well as 5250, say of a system had 200 5250 sessions running on it would it handle 200 ProfoundUI sessions, and the answer was yes, that was what they had observed.

This is not a solution that can be applied retroactively to older OS versions, but going forward every iseries should include ProfoundUI as the standard GUI interface. In a way IBM should buy them and make it official, but in another way these guys and the iseries community are probably better off letting them continue to innovate on their own. The engineering certainly couldn't get any better integrated with the iseries even by being acquired by IBM.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Thu May 06, 2010 5:07 pm    Post subject: Reply with quote

Henrik wrote:
Ralph,

I'm just curious ... how do they handle 5-10 sessions at the same time, how do they login and logoff - it is actually the big problem that nobody has cared to ask - you are among the - please ask ...


well, I know they run in QHTTPSVR, and I know just enough about that to be dangerous. I can only relate to what I've seen from CGIDEV2, some of the discussions you've been heavily involved in.

One aspect is the Apache HTTP persistent connection usage, which IIRC puts a keyword in the HTTP heading and Apache links to a consistent CGI job. IBM CGI doc recommends no more that four or five of these, and there are maximum settings in the Apache CGI server configuration, but to me it's esentially pretty much the same overhead as the same number of interactive sessions, but in QHTTPSVR instead of QINTER.

My guess is IBM didn't want to encourage RPG programmers to emulate interactive sessions with CGI jobs because that's the way we've always programmed, program per signon sitting there with state, that is all variables, file positioning, etc. left as is from last I/O and only same signon ever comes into program, which is completely different from non-persistent stateless CGI programming where everything starts over every I/O, any user coming into any instance of the program, and web serving programs having to be written accordingly.

Another aspect is the Apache HTTP configuration directive requiring iseries signon. I believe you posted about that extensively in terms of overhead, etc., and I do not know that ProfoundUI applies user profile security as with a signon or some other way. They specifically talked about security settings, role based access and such, and it may be entirely based on their own applied limitations to access menu options, programs, etc. with iseries profile not involved. I don't know.

I also do not recall specifically that they support multiple signons per user, I do know they have a session id, and I can imagine trying to support multiple session id's with cookies would be a problem but session id's as encrypted hidden fields I would think isn't a problem. If you open the same startup page in 10 tabs, each new tab would come in without a session and respond with a signon prompt and get a session id assigned and embedded in pages for that tab, so you would have 10 signons, 10 session id's, and 10 instances of the startup CGI program. Again I have no idea how ProfoundUI handles it exactly.

I also know that I am working on a personal project to demo multiple sessions with state from a dataq server program, so Apache HTTP persistent connection is not involved with that, and ProfoundUI could be doing something like that as well. Hopefully someone who knows a bit more can add some info.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Sat May 08, 2010 2:08 am    Post subject: Reply with quote

Jon Paris wrote:
As Mark Duignan of Lansa has pointed out, in my post of April 26th I over simplified the RPG needed to avoid dual maintenance. Of course RPG's requirement for unique format names precludes such a simple solution. Even using qualified names would require conditioning on each I/O operation.

But thinking about it reminded me that there is an even simpler option - use conditional compiler directives so that a single file definition can be used but with the HANDLER keyword made conditional. e.g.

Main F-spec entry

/If Defined(HandlerVersion)

HANDLER ....

/EndIf


You'll end up creating two program objects this way of course which would require some other corrective action (like separate libraries for example) but it can be done. My feeling is that there aren't enough real 5250s left in the world for this to be a real problem. Once people decide to switch they will switch everything I suspect.

Perhaps more to the point, one argument that could have been made wasn't. Namely that if you intersperse menus or non-RPG generated screens in the mix that RPG OA can't help with that. Both Look and Profound have solutions for this but it would have been a more valid criticism.

Of course it is quite likely that I've got this wrong too as it is not yet 11am and my brain does not fully engage until at least 2pm.


and a few cups of coffee, Jon :) It turns out it's even more complicated than that, or much simpler depending on how you look at it, at least with ProfoundUI (with tradeoff being high quality output for intended display). I will add a bit more detail on this in the forum thread.

The WRKSTN DDS is generated by the ProfoundUI (ProUI) web page Designer program. It has no actual DDS line and column output specifications. It won't display to a 5250, and a 5250 DSPF DDS won't be handled by the ProUI handler.

So in an RPG program that could display to either, you'd have two different WRKSTN F specs and record formats, once built by Designer based on the Designer web page layout which would have the HANDLER keyword, and the other being the original or a new one created for 5250. Both would have USROPN statements on it.

A decision must be made in the RPG program as to what the user display is, and the appropriate WRKSTN F spec opened. I guess checking if job is running in QINTER is clue to open 5250 WRKSTN F spec, although I don't know offhand how it gets submitted to QINTER versus a batch subsystem with no WRKSTN open yet.

In addition, at least for ProUI, web pages designed for mobile devices or any other different layout will be designed separately and have its own DSPF DDS, and RPG program must determine output device and open appropriate WRKSTN F spec.

Obviously any automatic handling of normal web pages for mobile devices and such is beyond scope of both the RPG program and the ProUI handler. ProUI handler would render normal web page and what happens to it after that is out of its control.

But to have different layouts for different devices would require a WRKSTN F spec DDS record format for each one as created by Designer when you lay it out, and only the appropriate one opened by the RPG program based on some external criteria.

I don't know how others are handling that aspect of it, but if mapping a set layout, say the original 5250, to different layouts at the handler level, probably only able to work within 5250 constraints. As I'll add to the forum thread, ProUI is not constrained by 5250 limitations.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
rd



Joined: 13 Sep 2002
Posts: 9207
Location: Jacksonville, FL

PostPosted: Mon May 10, 2010 12:47 am    Post subject: Reply with quote

Responding to Dan Burger's column at IT Jungle:
http://www.itjungle.com/tfh/tfh051010-story02.html
Open Access for RPG Grabs Attention at COMMON
by Dan Burger
May 10, 2010


Hi Dan,

A good column you wrote. I did not attend the RPG OA session as I had already read the IBM specifications for it and it's pretty straightforward. I did attend the ProfoundUI session and listened carefully to BCD's point of view at their booth.

BCD is a 4GL, so going with that or other 4GL / CASE products is not comparable to extending RPG as RPG OA does. It's not a better or worse comparison, just totally different strategic directions.

As far as not getting RPG OA, I agree and my thoughts from the beginning are that the enablement IBM provides for RPG OA is pretty trivial, especially for charging a run time fee for it. They internally in the RPG compiler replace an EXFMT (or any other IO command to an F spec with the HANDLER keyword) with a call to the HANDLER program specified passing the F spec data.

Anyone can comment out an EXFMT and do the same thing. The problem wasn't doing that, the problem was that there wasn't a standard GUI interface program to call.

On the other hand, it legitimizes a standard transparent way to take familiar IO and transparently invoke third party or custom writtem handler programs written in any ILE language, no reason why it wouldn't be written in RPG. You're doing API calls at that point, sockets and / or HTTP, and other i/OS API's and calls to the hamdler's third party library programs if it's from a third party. In other words, same stuff we've always done such as with CGIDEV2.

The state stuff from IBM saying the handler supplies state I don't understand. It may be just the way they're saying it, with a marketing slant to it, but is incorrect as far as I know. We could not use familiar programming if state wasn't preserved in the programs in variables, file pointers, etc. if the handler was preserving state. It would have to somehow refresh the program to a previous state from a stored session, and it doesn't do that. It just takes data and returns it, as with 5250.

All in all, my opinion is that RPG Open Access and ProfoundUI should be considered a standard addon to an iseries, in the same class as Robot and DBU, considered part of the total cost, and used in lieu of 5250 where the extra effort for richer screens is justified.

This is engineered in such a fundamental way to what we do that it should be considered a standard GUI interface to the iseries, even though any number of other GUI solutions are also available to be used.

rd
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Display posts from previous:   
Post new topic   Reply to topic    www.justiceforchandra.com Forum Index -> www.rdwrites.com forum (part of www.justiceforchandra.com) All times are GMT - 4 Hours
Goto page Previous  1, 2, 3, ... 10, 11, 12  Next
Page 2 of 12

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group