Advantage of Performance Modeler for TM1 Expert

[kukuxumusu]

Hi

I write this post because I didn't find any discussion about this subject in this forum and this question haunts me. I don't want to miss something but I don't have enough time to investigate myself.

I wonder what's the advantage of using PM for a tm1 expert... Working with TM1 for years, I am very confortable with architect. I have build several "tools" (essentially generic and reusable TI processes) that help me to go faster and I can always do what I need to do.

I trained on Cognos PM but I did'nt get what interest I have to invest more time on it... I understand that it is a geat "commercial" tool, than may give the impression that anyone can develop an application easily . But as far as I'm concerned I don't feel confident with it because I not sure of what he is doing in background. So I have the feeling of not control anything. My feeling is that he is very "chatty" (creating a lot of objects as TI process in all directions that could become difficult to manage) and doesn't cover the range of possibilites offered by architect. I'm not sure that PM will be able to do a very fine tuning on feeders for example. And I would not be surprised that one should then call an expert to correct an application developed with PM. But maybe am I all wrong!

To give a little context, I would say than my projects are made of 60% TI process, 15% rules and 25% user interface.

So my question is : are there TM1 experts using it and why? In what situation?

Thanks for your opinion and feedback.

Aude

[Alan Kirk]

I trained on Cognos PM but I did'nt get what interest I have to invest more time on it... I understand that it is a geat "commercial" tool, than may give the impression that anyone can develop an application easily . But as far as I'm concerned I don't feel confident with it because I not sure of what he is doing in background. So I have the feeling of not control anything.

That pretty much summarises the Cognos design mindset. "Don't you worry your pretty little head about it, our code wizards will take care of all of that. So you say that the resulting objects, rules and feeders are cr@ppily designed? No, can't be. After all, one size fits all!"

My feeling is that he is very "chatty" (creating a lot of objects as TI process in all directions that could become difficult to manage) and doesn't cover the range of possibilites offered by architect. I'm not sure that PM will be able to do a very fine tuning on feeders for example. And I would not be surprised that one should then call an expert to correct an application developed with PM. But maybe am I all wrong!

You're not. You're so very, very not. In fact I would say that you are about as on target as a laser-guided smart bomb.

To give a little context, I would say than my projects are made of 60% TI process, 15% rules and 25% user interface.

So my question is : are there TM1 experts using it and why? In what situation?

I might consider using it at gunpoint, and even then:
(a) It would take many, many guns being pointed at me. Indeed, we're talking a "final scene of Butch Cassidy and the Sundance Kid" number of guns pointed at me; and
(b) Even then there would still be only a 50/50 chance that I'd agree to do it, rather than still take my chances with the TI Editorsaurus.

When you say that 60% is TI, well, for mine I made my views about the Performance Muddler's TI editor pretty clear here. It's ridiculously slow to fire up, whereas the TI Editorsaurius is right there. It's right in front of you. It's ready to go with zero time delay in getting all of the Java code up and running (or not, as the case may be). The editor in Muddler does have some new features, features which have been implemented (in my view) very badly and in a half @r$ed manner. Everything from the woefulness of the find and replace functionality to the utterly awful "almost but not quite, in fact not even close" attempt at "something kinda-sorta like Intellisense but without most of the useful features" is just badly done in my view.

The Editorsaurus is in no way acceptable, let's be clear about that. A text editor interface which doesn't allow formatting, doesn't allow find and replace... not in any way acceptable in 2013. The lack of find and replace was barely acceptable in 1999. But for mine Performance Muddler's TI editor is little better in terms of overall productivity boosting, and if I need to do intensive text editing in a process I typically use Notepad++ with the user defined TM1 language. That's still more useful overall than the Muddler editor is, again in my view.

If you want to see a TI editor done right, then the place to look is Cubewise's Vizier. If IBM had any sense they'd just licence the thing except, of course, that Vizier was (I believe) written in .Net and of course that doesn't fit with the anti-Microsoft jihad exemplified by the attempt to move away from Excel and to Java with its famous "Write Once, Fail Everywhere" and "New Day, New Security Leak" code paradigms.

The reason that I am somewhat strident in my views on Muddler is pretty simple. The shortcomings of the TI Editorsaurus have been known for years. The extent to which they cost us productive time has been known for years. Over time, the extra minute here, extra minute there that the obsolete Editorsaurus costs us becomes hours. The hours become days. I hate to think how much time I've wasted tracing down syntax errors just because on a modern display the text size is almost a microdot. How many compile errors I've had because I forgot to replace one expression. How often I've had to shift code out to a real editor, then move it back in to compile it. How many times I've had to copy the variable names for a CellGetN formula and then fluff around formatting them and putting commas between them. How much time it has cost me to change variables from Ignore to Other since you can't set a default. Seconds to minutes, minutes to hours, hours to days, it goes on and on and on. People would talk about it in forums, at user conferences, anywhere where one or more shall gather in the name of TM1, verily this shall be discussed. It's not like IBM could be unaware of what practical but unsexy things we wanted in the way of productivity enhancements.

But instead, we got that {points with distaste to Muddler's ugly browser URL}. Something which seems to have been designed from the ground up to be superficially appealing in a sales presentation where you get to see brightly coloured icons which look the part, as long as you don't look too hard at the underlying functionality. And as long as you don't register how similar some of the icons are either, making it a GUI design that I'd certainly have no pride in putting my name to.

They knew what we need to save us time and make us more productive, they didn't deliver, and instead they gave us something that is all show and no substance and that, gentle reader, is what gets up my nose about Performance Muddler all the way to the back of the nasal cavity and why my pronouncements on it are rather less than mellow-tempered.

So in answer to your question... I cannot see myself using Performance Muddler in the near future. It may be of some value for completely new admins (if teaching them bad coding habits can be considered "value"), but I suspect that more experienced users are better off continuing to stick their hands directly under the hood.

[jim wood]

Performance muddler is aimed at those migrating over from Enterprise Planning. (Adaytum) It looks good in a demo and as far as I know that's pretty much where it stays. Once you get past the demo all serious development happens within Architect. (Shorter than Alan's response I know, but I hope I just about covered it.)

[George Regateiro]

I tried to use it just for the object migration options that are available through it and even those were very unpredictable. Moving a dimension that was not attached to anything triggered a recalc of all the feeders in one environment, but in an identical model it did not. We had other weird instances related to the migrations.

Right now it is a solution that is trying to address too many problems and in the end does none of them (at least that I have seen) well.

[Alan Kirk]

Once you get past the demo all serious development happens within Architect. (Shorter than Alan's response I know, but I hope I just about covered it.)

Yours is a good "overview from the satellite". Mine was more "from the house-to-house fighting on the ground" perspective with special emphasis on the TI editor since Aude indicated that that's where 60% of the time goes. (And since so much of my own time goes on TI, which means that it hacks me off above all else in Muddler.) But if you want short, I can encapsulate into four little words. Here we go... {clears throat, cracks knuckles, adjusts bow tie on the tux, leans in to microphone...}

They screwed us. Again.

George raises another relevant point with regard to hot promotion, a long overdue feature which shows no sign of being added to Perspectives / Architect. ( Indeed I don't think that there has been a great deal of work on Perspectives / Architect since the addition of Applications (no, the other "applications") in 8.3 aside from the obligatory Fresh! New! Look! changes of icons every version which instantly make all training material obsolete. Our material, not theirs, since their newer documents don't contain screenshots any more. )

Right now it is a solution that is trying to address too many problems and in the end does none of them (at least that I have seen) well.

Well said. Addressing many problems in one package isn't necessarily a vice; the aforementioned Vizier covered a number of gaps in TM1 functionality with the notable difference between Muddler and Vizier being that (from what I can recollect in my limited time playing with it[1]) Vizier mostly worked as advertised. On the one occasion that I tried the "transfer out" functionality in Muddler, though, it ended in Java errors and a very great deal of distinctly acidic profanity. I know that I should have attempted to reproduce the error and report it but by then my frustration level had redlined and I decided it would not be in the best interests of my health to pursue a further relationship with Muddler at that time.

-------
[1] So why am I not a Vizier user? No way would I be able to justify the cost given that some of what it does, I already had home written tools to do. Before I could stump up the money I'd need to justify the cost in time and money and I couldn't do that. Bear in mind what I said about the insidious nature of how the TI Editorsaurus robs you of time; a bit here, a bit there. It's only when you've worked with it long enough that it builds up. Consequently take any comments that I make on Vizier with the grain of salt that I only got to play with it for demo purposes for a shortish time. It still seemed to me to be better designed and executed than Muddler is, as best I recollect it.

[me2]

Alan Kirk wrote:
How much time it has cost me to change variables from Ignore to Other since you can't set a default.

You probably already know this but just in case: you can copy the Content setting from one variable and then paste it to multiple other variables in one go. Not as convenient as choosing your own default but it may save a few precious seconds nevertheless...

[Wim Gielis]

Alan Kirk wrote:
How much time it has cost me to change variables from Ignore to Other since you can't set a default.

You probably already know this but just in case: you can copy the Content setting from one variable and then paste it to multiple other variables in one go. Not as convenient as choosing your own default but it may save a few precious seconds nevertheless...

Or... select multiple boxes for Content, and change the Ignore to Other for the box that is selected but in white. See attached file.
I Ctrl-clicked every other variable, and now applying the change for the 9th variable will change all selected variables.
Hence, you can also select the entire column and change the Contents settings for all variables at once.

This method also works for the Variable Type column (String or Numeric).
You would think this method also works in the Advanced > Parameters tab (the type being String or Numeric). Nada...!!! That would have been too easy.

Wim

[Alan Kirk]

Alan Kirk wrote:
How much time it has cost me to change variables from Ignore to Other since you can't set a default.

You probably already know this but just in case: you can copy the Content setting from one variable and then paste it to multiple other variables in one go. Not as convenient as choosing your own default but it may save a few precious seconds nevertheless...

Or... select multiple boxes for Content, and change the Ignore to Other for the box that is selected but in white. See attached file.
I Ctrl-clicked every other variable, and now applying the change for the 9th variable will change all selected variables.
Hence, you can also select the entire column and change the Contents settings for all variables at once.

I knew about the one mentioned by me2, but have to admit that I was unaware of Wim's one; that's probably because I have copying and pasting the whole lot down as muscle memory now. (Change the first one (and don't I love it that "Other" is juuuust below the bottom of the visible items on the drop down so you have to scroll to it), click, [Ctrl]+[C] twice just to make sure that it's picked up, hold down [Shift], select the last field, [Ctrl]+[V].) I'll start trying to develop muscle memory to use Wim's one since it's fractionally faster and since we're not likely to see any functionally useful developments in the Editorsaurus in this geological epoch.

The thing of course is that I have never once needed to have a variable as Ignore. Sure, I do ignore some of them frequently, but I've never needed the contents type to be Ignore. Accordingly I generally change the whole lot to "Other" using the Copy / Paste method described above. You would think that they would all be left on "Other" by default. But no, because of course (a) the auto generated code selections must take priority and (b) nobody in Iboglix seems to have done any work on the Editorsaurus since the year started with a "1" rather than a "2".

[kukuxumusu]

Sorry, I didn't get back earlier. I simply put the cat among the pigeons and ran away! I just read your answers and especially that of Kirk, and I laughed fit to burst!!! How can you described my experience so well!

Everything is true. Special mention to the training material as I just updated mine from 9.5 to 10 a few weeks ago : very highly intellectual activity… About the icons you didn't mention that they are also indescribable : try to explain to a user where he should click on the subset editor toolbar by phone! ("the blue one with an arrow… no, the other blue one with an arrow …???!!!"). The best of the best are the new snapshot and slice icons in 10.

In my case I use Wim method for variable content (select all variable and changing the one in blank). I'm so "brainwashed" than I didn't even think it could be the default value without the autogenerated code that nobody uses.

About wasting time in TI process, I'm surprised you did not mention the copy/paste shortcut. Even when you know it you fall into the trap once two! And the best part is when you do not notice immediatly...

I discovered the notepad++ TM1 language add-on a few years ago and it changed my life, my eyes are better since that day! I wonder how long I have NOT lost with it in comparison to TI editor.

Well thanks a lot for your feedback that confirms my idea. I didn't want to be the only one who miss something like old persons who don't want to change Well, to resume "nothing to see, move along".

Thanks also to IBM to give us a new problem : explain the client at project start that "no we won't use this wonderful tool that you saw in the pre-sale demo" and argue about it to justify.

[jyoung66]

While I agree with all that has been written above, there is the inescapable fact that even the experts will have to get their head around PM and soon.

10.2 is heavily PM centric, and IBM have made it clear that this is the shape of things to come

Clients of the experts will be developing in PM and calling us in to fix it when it breaks or to enhance their builds when it doesn't quite do what they want it to.

[lotsaram]

While I agree with all that has been written above, there is the inescapable fact that even the experts will have to get their head around PM and soon.

10.2 is heavily PM centric, and IBM have made it clear that this is the shape of things to come

Clients of the experts will be developing in PM and calling us in to fix it when it breaks or to enhance their builds when it doesn't quite do what they want it to.

A valid point but I dread the day. Undoubtedly I'll give it a go again with 10.2 but experiences with 10.1 were not good. The eclipse client is simply a bloated pig that wastes untold time downloading meta data you simply don't need before the client is ready and available. If you have a "chunky" model of more than a few dozen cubes and 100 or so processes then you can comfortably go for a coffee waiting for PM to load, and not just get a coffee but actually leave the building and down an espresso from the local barista and make it back before PM is ready for action. Add to that object save and refresh time. The client is simply too slow and clunky to use as a development tool. If it wasn't so slow I would be using it for TI editing but I can't afford the lost productivity.

Mind you the new rule editing is quite awesome (compared to the shiite we have been putting up with) given drag and drop of attributes and functions and side by side with cross tab to see the results. But it really beats me why they had to go and invent another language syntax when surely one is enough. The parsing of one syntax to another is one extra thing for developers to learn and one additional break point. However the thing that really makes rule editing with PM untenable for me (more than the mind numbingly slow eclipse client) is the lack of control of rule statement order in the compiled rule file. The lack of top to bottom visibility of the whole rule file is a major miss given the criticality of order of area statements in determining what calculation will actually apply to a given cell intersection.

As a fan of integrated login the support for only TM1 authentication and CAM is a big irritant as is the sheer vulnerability to breaking when java updates come along (like they do every 5 minutes). After either a java or windows update recently the previously functioning PM client in our production environment now just returns a java error on launching. Nothing seems to be able to fix it and no a reinstall of the entire CX environment isn't something I'm contemplating when we can live perfectly well without it. Unfortunately it will ned to be fixed in order to be able to publish some applications we ave planned.

OK with the necessity of PM for contributor and central application building and the new kpi metric cubes. Achingly close with TI - just give me a client that loads in 5 seconds not 5 minutes. But for rules just give us something that can do everything the plain text editor can do and THEN add the fancy tricks. Cartwheels and loop the loops make no sense if you can't take off and land.

[Alan Kirk]

While I agree with all that has been written above, there is the inescapable fact that even the experts will have to get their head around PM and soon.

10.2 is heavily PM centric, and IBM have made it clear that this is the shape of things to come

Clients of the experts will be developing in PM and calling us in to fix it when it breaks or to enhance their builds when it doesn't quite do what they want it to.

Not all of the experts have clients, fortunately. Not because I have anything against clients, or sites which hire outside expertise. It's just that when you're in house, as some of the heavy hitters still are (though I admit a minority these days), you aren't limited to what Iboglix gives you.

I'm not sure that I'd agree with the "and soon" part, though, even for consultants. Just because something was developed in PM doesn't mean that it needs to be debugged in PM. The Perspectives toolset isn't going away for at least a coupla-few versions.

Speaking personally I'll get my head fully around it simply because it's currently my job to. Once I've done that I'll record a couple of TM1 Bytes on it (assuming that the amount of profanity gets through You Tube's filters), then refer back to those if I need a reminder of how any part of it works.

Then I'll use other tools day to day.

(I note that Lotsa posted a reply while I was typing mine. I think that he hit a lot of nails squarely on a lot of heads. I don't agree with his view of the TI Editor in PM; as I've gone into in some detail previously I think it is still lacking in a lot more respects than load time but I'm getting closer to a point where I need not worry about that again. But I think that the overall thrust of his post is spot on. IBM can push PM all they want but as long as it doesn't deliver (and it may one day, I don't dispute that), those with expertise will find ways to work around and through the thing.)

[Duncan P]

However the thing that really makes rule editing with PM untenable for me (more than the mind numbingly slow eclipse client) is the lack of control of rule statement order in the compiled rule file.

Both in 10.1 and in 10.2 PM you have the ability within a rule window to reorder the rules by pressing the rightmost button.

rule_reorder.png (12.4 KiB) Viewed 15748 times

Once they are reordered subsequent rule regeneration is supposed to respect the new order.