BrianL wrote:Just to add my $.02.
While I've never attempted to use the .NET api, it's been my understanding that the older .NET TM1Web was written entirely using it. Either I'm wrong about that, or the documentation is so horrible it's impossible for anyone outside IBM to do anything meaningful with it.
Indeed, it's entirely possible that at the very least they shared some DNA, possibly a lot of DNA. The .Net API was released with 9.1. Web was around long before that of course, but I recall that there was a barrier that was passed with 9.1 Web. (Web 9.0 and prior, effectively unusable without much head banging, Web 9.1 the first half way decent implementation. Not perfect, but genuinely usable. I recall that 9.1 was my first go at implementing Web and that we had looked at 9.0 and earlier at the time and said "nooooooooo.....")
That being the case, development on the .Net API and the 9.1 and later IIS Web models would have been kicking along at the same time, so it would make sense if they were connected in that way.
BrianL wrote:Even if IBM decides to stop supporting the C/.NET/Java APIs, I can't imagine they'll stop working anytime soon. It'll probably just be a documentation/helpdesk issue. I say this, because the C, .NET, and Java APIs all send the same exact bits over the network. The .NET and Java APIs were more of an attempt to make the C api look more object oriented. From my brief perusal of those two APIs, they seem almost identical in object design. The only alternative API that doesn't use the same protocol is the new REST api, and even IBM's own software doesn't appear to be making extensive use of it yet.
Maybe
one exception...
BrianL wrote:Bottom line is that removing the capability for the TM1 server to accept C/.NET/Java API traffic will result in pretty much every client application in existence to stop working. This includes TM1Web, CAFE, Architect/Perspectives, Performance Modeller, Insight, etc...
That's actually a damn good point, with one qualification; at the most excellent (with the exception of
one presentation,
{cough}) Cubewise conference last week Ronnie Rich from IBM hammered the point that CAFE is powered by REST. This fits with what was already known; somewhere in the bowels of the forum there will be some posts from when CAFE first made an appearance. I recall that some guys who had been on the testing program mentioned that it was based on a "new API" that nobody knew much about at the time but which we now, in hindsight, know to have been REST. Also there's the fact that CAFE is the only non-Web client that doesn't perform like cr@p over a WAN, pretty strongly suggesting that it's powered by something different. As Ronnie said CAFE just wouldn't have been viable using what he referred to as the old C API.
But yes, no supported C API would probably mean no supported Perspectives / Architect. Ronnie did mention that there are obviously plans to "sunset" Perspectives / Architect, but it won't be until CAFE has about 95% of existing Perspectives functionality, which won't be for a while yet. Many users had indicated that the lack of Perspectives functionality (including formulas) was their biggest barrier to implementing CAFE. I'm not sure where the missing 5% will be (indeed I didn't get the impression that there was a definitive plan for that yet) but I imagine that it would be on the administrative side. I got the impression that most administrative work would pass to Prism. (Which was confirmed as being only a codename. Actually I don't think it's a bad
name in view of what it does, just that it has bad associations that a data services company would do well to distance itself from. But I digress.)
We were told that the ultimate goal is to reduce the plethora of UIs down to two, CAFE for the Excel side and Prism for the web side. I wouldn't be wildly enthused to lose (for example) .xdi and .xru ability
if that moves out of Excel and into Prism. I hold fast to both of those (unless the updates need to be automated by data feed) for the reasons that I've often stated in the past. (Ability to format clearly, ability to use Excel formulas (especially in Rules) to compile expressions, ability to store meta information about the dimensions in the sheets, ability to store prior versions in a workbook.) But even if that were to happen the RESTful API seems to offer the potential to knock up a bit of code to still use workbooks (or any other type of document) as a source anyway.
And at this point we've managed to segue the thread again more into "Future TM1" rather than just "Future TM1 APIs", but no matter, that happens...
To be honest after so much frustration about the UIs and APIs, particularly with hateful cr@p like Performance Muddler and the exceptionally detestable Cognos Configuration for TM1, and the irritation of the (apparent) swing away from .Net toward Java, I'm actually feeling if not (yet) wholly optimistic, certainly leaning strongly that way about where the product is headed in this respect. (I don't think I'll ever get to like the lumbering overhead that is CAM security, but generally, I mean.) Even the Tomcat TM1 Web; yes, it still has a stack of bugs as you would expect with any de facto v1.0 application, but at least (a) it's nice to
finally have websheets that actually look right on screen rather than the "throw of a dice" column width arrangement with the old Excel-powered Web application and (b) it will be nice not to have an ocean of orphaned Excel sessions left running on the server. That reminds me, how many do I have on 9.5.2 Web this morning? Ah, 11. Lovely. Kill, kill, kill and more kill. The memory bloat on Tomcat web is less great, but on the whole I've swung my opinion around about whether
on balance, taking into account all of the pluses and minuses, this change was a good thing.
The Cubewise Conference was great for getting a handle on this "big picture" stuff, and if next year's has a guest list anything like this year's (with one exception
{cough}) it's definitely something that I'll be most looking forward to attending (and would definitely encourage anyone else near one to do as well). The more of the community that gets along to events like this, the better engagement we'll get with IBM.
Oh, and also... Ronnie confirmed that
TM1 Top is returning in fix pack 4, woo hoo!
Granted we don't use TM1 Top as actively as we once did now that we have Pulse, but Pulse is tied by licence files to specific servers and for ad hoc (usually virtual) ones that I spool up for version testing I know that I much prefer nice, light TM1 Top as my monitor of choice rather than firing up Operations Console.
"Hmm, I need to know which threads are taking so long. Let me just fire up Operations Console... ah, I have time for a quick movie while it loads."
{Pulls out Titanic, The Director's Cut, reaches the point where Rose promises to "never let go", then lets go, does the "forward and reverse" thing...}
"Leo goes down... Leo comes up... Leo goes down.... Leo comes up.... Oh look, Ops Console is up!"
Yeah, I think I'll be happy to see TM1 Top again.
May be one day..... But lucky enough to meet another TM1 legend in person while ago!
