Experience with PAW and its future prospects

[amit_hhh]

Hi Everyone,

I want to get an insight on how the experience with PAW has been and what are the challenges that we still face? Do you think that we can completely get away with TM1 architect for development and adopt PAW? What is its future prospects? Thanks!

[Elessar]

Hello!

IMHO:

• PAW has many disadvantages, but the Hierarchies, which you cannot use in Architect, outweigh them all.
• Rule/TI editors are, I can say, "handy" now. Ctrl+space is smart enough now to let me write a rule without touching a mouse.
• You can create really useful dashboards: with input cubes, graphs and buttons in one screen (not as flexible as in TM1 Perspectives, though...)

[Steve Rowe]

I'd say each of the individual components of the PAW development area are better than the Architect flavour but..

When I am developing I may have 2 TIs, a rule sheet and two cube views visible all at the same time and I need to swap between them all or open up another object without breaking my flow.
For me architect is better at this by a distance, now some of this is muscle memory but alot is due to
A. The size of the objects in PAW (i.e. text size and real estate consumption), this means I have less space to work with.
B. The ease and speed at which you can drag and rearrange the widget panes.

For new developers you should be fine in the new tool.

[Wim Gielis]

Indeed, for experienced TM1 developers, the efficiency goes down when developing in PAW.

To name just 1 thing, if you press the first letter of a cube name you would jump to the first cube whose name matches the letter. That is what happens in Architect. Don't try this in Workspace, you need to scroll all the way down.

There are still a number of things that need to be implemented, for instance, grabbing TI process parameters from the canvas holding the process button.

PAFE can often be a royal pain in the backside, for no apparent reason, certainly if you drag a PAFE report onto a PAW canvas. Likewise, dragging a Perspectives report onto such a canvas in PAW, it often does not behave the way you intend to. Or PAW is painfully slow and it takes 5 seconds to respond to a mouse cursor change.

For end users, overall it's an improvement, though.

[declanr]

Do you think that we can completely get away with TM1 architect for development and adopt PAW? What is its future prospects? Thanks!

I will answer this question purely from the development point of view that you have asked.

I honestly think that I could. I do the majority of my development directly in the PAW interface now; every so often in a panic I still jump back to architect but that's because I worked with it for so long and its a hard habit to break on occasion.
I feel like every time that I say I enjoy working with PAW (or PaFE) here it sometimes seems like I am committing blasphemy to some people. But the truth is I prefer it to architect.
I found it surprisingly hard to get over the muscle memory of doing things the way that I used to do them but that was my own stubborn brain fighting me on it.
There were certainly things in the early days that were real problems (like setting variables in the TI editor which was solved a while ago now) but every release (of which they were/are frequent) improved on it and I am confident it will just keep improving.

When I am developing I may have 2 TIs, a rule sheet and two cube views visible all at the same time and I need to swap between them all or open up another object without breaking my flow.

5 objects to be working with while fully engaged at the same time is a lot. I do have the benefit of a large screen but when developing I often tend to work with the 3 section predefined page (full size left side of the page and right side split into 2) and I would have say a TI or Rule on the larger left hand pane while having 2 cube views on the right hand sides... in this example I would be moving data between Cube A and B while swearing at the stupid typos in my TI code.

There is also nothing stopping me from having another PAW book open in a different browser tab on my second monitor - where I can have the other objects that I do need to reference but don't need to look at constantly.

A. The size of the objects in PAW (i.e. text size and real estate consumption), this means I have less space to work with.

For cube views you can reduce the font size on the properties pane.
For TI and Rules editors you can reduce the font size by clicking "Process Editor Options" or "Business Rules Options" (the 3 dots in the upper right corners.)
The default size is 14 which can seem huge - I tend to reduce it to about 10 for my own personal preference.

For new developers you should be fine in the new tool.

I could not agree with this fact more. I worked on a project where the customer was upgrading to have Workspace available relatively early on in the development but we started in 10.2.2. In the project I was training a couple of in-house super users to learn some of the basics so that they could support after it went live. They got to see architect for about a month and then had workspace available. They picked it up so much quicker in Workspace.
When we have worked in Architect for a long time; we sort of forget about all of its problems because we subconsciously know the way around them and had nothing/little else to compare it to.

https://ibb.co/NpmS4S8 <-- Example screengrab of some sample system with 3 windows on size 10 font.

[paulsimon]

Hi

I would agree with the comment about the usefulness of Named Hierarchies. However, I use PAX for that rather than PAW because I find it easier to use, apart from the issue of half the buttons not showing in the Set Editor, which is still an issue in 2.0.9. Some people have this issue and some don't. IBM don't seem that interested in getting to the bottom of why.

I still use Perspectives for rule development and TIs. I only use PAX to check for any issues in the generated hierarchies. However, I now have a set of standard processes for maintaining hierarchies so haven't used PAX for a while.

We are still having issues with SSL certificates and until those are resolved we can't use PAX or PAW off the server so we haven't as yet pushed it to end users. Admittedly we have more complex networking than most.

I will at some point try the development side of PAW but quite frankly in our current 2.0.5 version it seems unreliable.

To expand on a previous comment
"There are still a number of things that need to be implemented, for instance, grabbing TI process parameters from the canvas holding the process button."
Please see my earlier post on this https://www.tm1forum.com/viewtopic.php?f=3&t=15318 and to vote for IBM to fix this https://ibm-data-and-ai.ideas.aha.io/id ... itted_idea

In my view until you can link the parameters of an Action button to selections on the PAW book, then PAW is useless for building applications. It is only a reporting tool. Planning Analytics is called that because users want an application that leads them through the planning process, and which manages workflow, runs process then jumps to screens, etc, all based on what they have selected on the screen.

The only way to do that in PAW at the moment is to embed TM1 Websheets, and if you have to do that you might as well do everything in TM1 Web. In fact if you do embed a TM1 Web sheet which navigates to another TM1 Websheet then PAW rather unhelpfully just changes the viewport over the TM1 Web sheet and you then cannot use the back button or access the tabs that TM1 Web will create for you. It is therefore not as good as TM1 Web.

The other issue with PAW is the need to maintain two sets of security.

I find the ability to push things from the cube viewer to Excel easily in Perspectives and to a lesser extent in PAX, to be very useful during development and adhoc querying.

I am not actually involved in our PAW/PAX evaluation as I spend most of my time these days working in JavaScript and SQL. However, it seems that there are additional problems with Sheets not converting to PAX correctly, which we are working through.

Once we get our various issues with SSL certificates, firewalls, docker, etc, resolved, and once we complete our evaluation we will make a decision on PAW. I suspect that for development in the future, we will have little choice but to use it. However, unless IBM resolve its many issues around application development we will look to buy Aplicqo or Cockpit.

Regards

Paul Simon

[ykud]

In my view until you can link the parameters of an Action button to selections on the PAW book, then PAW is useless for building applications. It is only a reporting tool. Planning Analytics is called that because users want an application that leads them through the planning process, and which manages workflow, runs process then jumps to screens, etc, all based on what they have selected on the screen.

The only way to do that in PAW at the moment is to embed TM1 Websheets, and if you have to do that you might as well do everything in TM1 Web. In fact if you do embed a TM1 Web sheet which navigates to another TM1 Websheet then PAW rather unhelpfully just changes the viewport over the TM1 Web sheet and you then cannot use the back button or access the tabs that TM1 Web will create for you. It is therefore not as good as TM1 Web.

The other issue with PAW is the need to maintain two sets of security.

This + 100.
PaW is also opening a TM1Web user session for every embedded websheet in a book, so if you go through 5 books you create 5 Tm1web sessions. This leads to even a moderate number of busy users in PaW generating a large number of Tm1 sessions if websheets are embedded. But 'it's by design'

[David Usherwood]

Spending some time developing systems with 'another' cube engine which mandates web for development, I worked round the issues by opening multiple connections in different tabs which I could tear off and rearrange. Main gotcha was that changes in one session would not show up in another without doing a browser refresh. I haven't tried this with PAW as I am quite happy with Architect/Perspectives until IBM finishes (hah!) making formula based reports, being the highest quality output, work with named hierarchies.

[scrumthing]

The other issue with PAW is the need to maintain two sets of security.

For me that is the worst part. It is AFAIK still not possible to delete users in PAW in an orderly fashion. At least not if you want to the ability for every TM1 user to simply log in. Especially if you want to host different applications in one environment with different instances and different admins it gets quite impossible.

I like the potential of the administrative agent but again no proper administration there. At the moment my favourite way of interacting with PAWAA is the rest api of the service. That looks quite promising and eliminates some of the painful powershell to get server information like RAM usage.

In my experience you constantly have to restart PAW as a whole because there is no proper debugging which service might have stopped working.

And of course the usual stuff like it is not usable for standardised reporting! Most of the „cool“ widgets are quite useless. The UI is just painful and is not at all customisable.

At the moment I would stick with TM1 Web and Perspectives or evaluate Apliqo. Or if both of them are for some reason not an option check out the good old Cubeware Cockpit. It is a bit outdated and ugly but it gets many, many jobs done. At least you hit the dead end a lot later than in PAW.

[macsir]

I forced myself to use more in PAW and now I get used to it. To me, the only drawback is you can't simply hit 'ctrl + double C' to copy names from PAW, which is convenient in Perspectives but it is OK. And it is quite convenient to compare codes between the environments via PAW as you can simply copy everything from Script tab into a comparing tool.
We use Linux version of PAW and it is fast and smooth. I never reboot it unless I made some configuration changes.
Next step would be using more in PAx.

[Alan Kirk]

My loathing of everything to do with PAW is well known, but also not germane to the issue. I have Arc. My company wouldn't pay for it so I pay for it out of my own little pocket. Oh gee, it works without Docker and connection strings and running endless command line scripts. And it works really, REALLY well. Also, while it is constantly evolving, it works without giving the feeling of being a half baked work in progress. There is no way that I can see myself EVER developing in PAW rather than in Arc.

[PlanningDev]

In my opinion PAW is better than what existed prior to it. What your end goal is definetly drives whether or not it will work for you.

If you want to stay inside the box and use the functionality it has then I do think it provides a better user experience overall. I like many have forced myself to use PAW and PAX in order to make sure rolling it out to end users would be a viable solution. I rarely go back to Architect/Perspectives now. Even if you have to use websheets you and feel like you should just stay in TM1 Web, you may as well bring them into PAW and potentially gain any benefit PAW may bring.

However, If your are capable and your end game is a custom web application then definetly PAW will fall short and Appliqo/Canvas etc need to be looked at. ARC is also a very good tool for development and has some very nice features.


Advantages
Synchronization of widgets
Multiple widgets on a canvas
Easier to create guided experience
Overall cleaner front end for end users
Much more capable of self service for end users
Auto code completion in rules/Tis
PAW selectors are better (IE searching for an element)
Personal preference but Im starting to like the subset editor better
Double click drilling down on consolidated elements
Custom aggregation without going into the subset editor
Column sort capability built into cube viewers

Disadvantages
No way to pass parameters to TI's
Formatting is limited for font/colors (Although custom fonts and pallettes are now possible)
Charting is limited
No control over how many rows are rendered/paged
Unable to select control objects as source in a TI

[declanr]

Unable to select control objects as source in a TI

They have actually sorted this one now - not sure what release it came out in but you now have a drop down in the TI Data Source to swap between Control Objects or a Standard Object.

https://ibb.co/gMY8PDh

[PavoGa]

My loathing of everything to do with PAW is well known, but also not germane to the issue. I have Arc. My company wouldn't pay for it so I pay for it out of my own little pocket. Oh gee, it works without Docker and connection strings and running endless command line scripts. And it works really, REALLY well. Also, while it is constantly evolving, it works without giving the feeling of being a half baked work in progress. There is no way that I can see myself EVER developing in PAW rather than in Arc.

Been using ARC for about 45 days now and love it. A few things take some practice to get over "muscle memory" with Architect as declanr put it. We had the trial version, but looks like we've convinced the company to get it for us. We've gone AWS and that is pretty much forcing us to wean off of Architect. Slow is an understatement when using Architect with AWS.

But I still use Notepad++ to usually write code and then just paste into ARC. I do use the ARC editors far more than I did the Architect ones, but still have too many things in Notepad++ I use that are not available in ARC.

ARC is a pretty darn good product based on my experience so far.