TM1 Forum

An interview with Hubert Heijkers, chief architect for the TM1 server, at the TM1 conference in Sydney:

https://www.youtube.com/watch?v=tL3puB0AK74

Pro tip: Keep your eyes on the right hand side of the screen.

Also, an interview with Manny on the subject of innovation:

https://www.youtube.com/watch?v=7wk6Q1PWkZk

Thanks Alan, will wear my left eye patch to be sure.

gtonkin wrote: ↑Fri Aug 23, 2019 7:57 am Thanks Alan, will wear my left eye patch to be sure.

I look slightly better in this unlisted one, which is about the split topics of the Forum and about the future of AI. I need to wear black more: https://www.youtube.com/watch?v=-SqF78PjofA

Black clothing is apparently very slimming but I caution against black underwear.

Alan Kirk wrote: ↑Fri Aug 23, 2019 7:47 am Pro tip: Keep your eyes on the right hand side of the screen.

So, apparently you're not Mark Harmon's twin. Surprise, surprise. You are a pretty good interviewer though.

Thx for sharing!

I'm very curious to hear more about that "scripting language" in the coming years. In a way I hope that its not a new language but just regular javascript or python or something.

tomok wrote: ↑Fri Aug 23, 2019 12:32 pm

Alan Kirk wrote: ↑Fri Aug 23, 2019 7:47 am Pro tip: Keep your eyes on the right hand side of the screen.

So, apparently you're not Mark Harmon's twin. Surprise, surprise.

Add a few cm, take off more than a few kg, close enough.

tomok wrote: ↑Fri Aug 23, 2019 12:32 pm You are a pretty good interviewer though.

Kind of you to say!

Bakkone wrote: ↑Fri Aug 23, 2019 3:50 pm Thx for sharing!

I'm very curious to hear more about that "scripting language" in the coming years. In a way I hope that its not a new language but just regular javascript or python or something.

There isn't that much more to hear yet; as Hubert said, options are still being explored.

At a previous conference he asked what our response would be if TI was to be replaced by a language like Go. (NOTE: This was not a proposal, just a "sounding out".)

I can't recall what my reaction was at the time; I don't think I was that enthused but I'd never actually heard of Go at the time (or perhaps I had but it was way out on the periphery of my consciousness), which made me question how many other finance turned admin people would know it either. (Though I hasten to add; the question was "something like" Go, not necessarily Go itself.

Still, as he said quite categorically the plan wasn't to replace TI but to supplement it.

Javascript would be known by, I think, a reasonable number of admins if that was to be the pursued path.

Python's advantage would be that Marius Wirtz's TM1Py project is picking up a decent head of steam and is starting to create a good body of example code. My concern with Python itself though is the way it uses formatting (indentation) to define scope blocks rather than more conventional delimiters like Function / End Function keywords or braces. In 50 lines of code that's not a problem. In 500 lines of code that's a problem, especially in an interpreted language where "BOOM!" doesn't happen until runtime.

It's not something that we'll be seeing in the immediate future, though.

Sounds interesting. Personally I think they should find something that actually CAN replace the TI part. Otherwise we will have three different "languages".

But I always tend to think about things like Canvas and all these other things beeing developed on top of the api. As much as I love it. I always wonder about customers using other frontends lowers the threshold to just move onto a different platform altogether. Im sure there are peoplpe at cubewise putting together the final pieces to make Canvas its own planning product with a javascript/MongoDB backend instead of TM1. I guess switching to Python or JS in TM1 would really accelerate something like this.

It seems every month there is a new player that has thrown together a new EPM product with some open source JS stuff some DB in the backgorund.

Bakkone wrote: ↑Mon Aug 26, 2019 12:54 pm Sounds interesting. Personally I think they should find something that actually CAN replace the TI part. Otherwise we will have three different "languages".

I'm not sure that would be possible or even desirable. Hubert said that his intention was to provide an alternative to what people are currently doing on the Prolog and Epilog tabs.

However the present TI language is internal to the server. In data and metadata it's not just a scripting language reading a file line by line and writing it up, but it's also (for example) doing things like optimising the locking within the model and handling contention and, if necessary, rollback.

Bakkone wrote: ↑Mon Aug 26, 2019 12:54 pm But I always tend to think about things like Canvas and all these other things beeing developed on top of the api. As much as I love it. I always wonder about customers using other frontends lowers the threshold to just move onto a different platform altogether. Im sure there are peoplpe at cubewise putting together the final pieces to make Canvas its own planning product with a javascript/MongoDB backend instead of TM1.

I'd be less certain about that. The server is one of two things that has always been at the heart of TM1; its performance, power and flexibility, its ability to handle ragged hierarchies (AAAARGH! Don't say "Hierarchies"!) ragged consolidation trees with ease compared to many other solutions, the writeback ability and so on. It's not possible to just whack another DB on the back end and get all of that.

The other thing is of course Excel integration; the ability to quickly and flexibly create ad hoc reports that draw from multiple cubes and multiple servers. The ability to create upload sheets where you just create a block of formulas, point them to the values that you've dumped into Excel from elsewhere and hit F9, etc.

I'm not disparaging Canvas at all; I think it's an impressive product. (The very thought of PAW and how it has been implemented, on the other hand, has me reaching for the Imodium.) But regardless, not everything lends itself to a web based front end anyway. That's why I think that the reintegration of the TM1 and Cognos front end teams is a net negative for us out here in customer-land. The Cognos team seems to be only capable of thinking in terms of "Pretty! Dashboards! Ooooooh, Preeeeeety! Reports!" and that isn't where the heart of the power (or appeal) of TM1 has ever been. It's a sideshow, not the main event.

But they seem to be incapable of seeing that which is why {exasperated sigh} two years down the track we still, STILL don't have hierarchy-aware DBRW formulas and are still not seeing them on the horizon.

Bakkone wrote: ↑Mon Aug 26, 2019 12:54 pm But I always tend to think about things like Canvas and all these other things beeing developed on top of the api. As much as I love it. I always wonder about customers using other frontends lowers the threshold to just move onto a different platform altogether. Im sure there are peoplpe at cubewise putting together the final pieces to make Canvas its own planning product with a javascript/MongoDB backend instead of TM1.

Then I guess you would be mistaken because I can assure you that this is not happening, and moreover is not something that is likelly to either.

One of the core value propositions of TM1 is the modelling engine. And there is no other modelling engine quite like it. The whole point about all the tools that Cubewise and Apliqo build is they work with TM1 to make it better and create more value for users and developers.

lotsaram wrote: ↑Mon Aug 26, 2019 9:04 pm
Then I guess you would be mistaken because I can assure you that this is not happening, and moreover is not something that is likelly to either.

Very quick to deny this are we not...

Thanks for your thoughts and insights.