i have heard this IDE has lot of helpfull features Included, hasn’t any tried this one yet?
I’ve installed it on a test PC and poked around, but because of this:
- No support yet for Linux build
I’m not doing any real development with it yet. Once they fix that, then I’ll force myself to use Eclipse for a while so I can provide useful feedback.
I started to use the new IDE a few days ago. I must say the compilation time is fantastic (about 3 secs for a rather big project). The downside is that there is no makefile anymore (or other build scripts), so here goes the linux support…
One big problem I still have is that I want to use the math library. Before I just had to integrate the libm.a, but this doesn’t seem to be so easy in the new IDE.
There are fields in project properties ->C/C++ build -> settings ->ARM ELF GCC C Linker -> Libraries. I’ve tried to set up different entries there with no success so far.
Not only Linux support, but also support for all developers who have a strong preference for their editor, which happen not to be Eclipse. I mean, it’s a well established fact that developers are very passionate about their IDE, and embedded developers probably even more so
But I can’t imagine that we would be kept without support for “real”, developer-friendly Makefiles; that would reduce the development environment’s usability too dramatically. And even if that issue was somehow overlooked, some feedback gathering must be planned about the new IDE, and this point would be bound to be raised quite fast.
Note that being based on Eclpise’s CDT, the new IDE uses the MBS building system, which is portable to linux; its “only” drawbacks are to be very complex, and incompatible with the universal Makefile standard…
To be honest, I wonder if this is true (unless “feedback gathering” means “wait for customers to yell at their distributors”).
- When I signed up for the IDE beta, I filled out a questionare. Once the beta was available, it was obvious that at the time we were filling out the questionare, they were already deep into development. So our answers to the questionare would have had little or no effect on their development plans.
- It took them weeks to add this new forum to support the new IDE. If they were eager for our opinions, wouldn’t they have created it sooner?
- The forum description says the IDE will be released “in early December”. So it seems they’ve already planned on releasing, with or without our acceptance of it.
As a Wavecom representative, please allow me to take these points in turn:
- Wavecom was extremely grateful for the feedback on the questionnaire and yes, the development had already started but to say we did not listen is simply not true. Firstly, there was an immediate change in some of the backlog priorities for usability. Secondly, we all must remember that software is a continuous process and we have a roadmap now beyond the first release and all the feedback in the questionnaire was carefully analysed to build this roadmap.
- The reason the forum was not added sooner was because we were in a controlled environment phase for beta testers. A dedicated tool was set up for that but almost no one used it. It would be useful to know why so we can improve the next beta program.
- The release is indeed due in December and as I said in point 1 above, much has changed in the final release compared to what was in the beta release. We’re not saying there is no room for further improvement but I cannot accept that yours and other developer’s valuable comments were not taken on board as this simply isn’t the case. The whole M2M Studio development team has worked extremely hard to get to where we are today, all in less than 9 months. I hope the commercial delivery will demonstrate this and I look forward to seeing fruitful discussions on the release and the next steps for M2M Studio. It is after all for you, not for Wavecom
Thank you for your response. Part of the reason for my previous email was to see if I could get a response out of Wavecom at all. It took a while, but at least now I know that Wavecom does eventually read the forums.
To be fair, I admit that Wavecom could have used information from the questionare to influence development and we can’t expect all our requests to be implemented in the initial December release.
However, all of the posts in the thread pretty much say the same thing: “I don’t want an IDE plugin. I want a simple makefile.” (OK, my first post didn’t state it, but my posts in other threads makes my opinon pretty clear).
This is the source of my complaint about development starting before the questionare. No matter how much feedback we give, the new “M2M Studio” is going to be an IDE plugin.
Flyspray is a bug tracking tool. If you read this thread, the posts are:
- A general question: “hasn’t any tried this one yet?”
- Multiple comments about a documented, unfinished feature: “No support yet for Linux build”
- A usage question: “I want to use the math library”
- And a comment about the general design desision to force users to use Eclipse.
None of these are bugs. Therefore, the but tracker doesn’t get used.
I think a bug tracker is a good idea, but since M2M Studio is quite a bit different from the previous IDE, we need time to understand how to use it before we will dig deep enough to find any bugs.
As For Jamie’s point 2):
I think a subforum should have been opened for M2Mstudio beta where only beta users would have had access.
and the responcetime could be improved. not especially for answers, but an indication that it is noted/read would help.
You can also add improvements in FlySpray.
Notice that in theory, the need for Makefiles is mostly orthogonal to the M2M Studio issue: as long as sane makefiles are provided, no Makefile-loving developer is going to mind whether there’s also a plugin for some editor in a corner of the CD. The fear that motivates the posts, here and in a couple of other SDK-related threads, sums up to: “We hope that Wavecom won’t believe that M2M Studio renders the need for sane makefiles any less urgent!”
First I’d like to agree with Matt with all the points he mentioned. After I used the IDE some more, I must say that it is a big step on the way to an integrated developing tool. But I would like to make two bold statements:
-the average user tests his new code directly on a module without using rte, in three simple steps: compiling, downloading the binaries, checking the text output in a terminal
-these three steps should be a one click function in the IDE, since they are performed over and over and should (if automated) certainly not take more than 20-30 seconds depending on the complexity of the code
The compiling is great in the IDE, the downloading is a bit difficult to reach and the integrated terminal certainly lacks customization options.
Thus I think that the IDE is nice for beginners, but not good enough for professional users. And for the latter there is an alternative. Basically you only need to have 3 key ingredients for a successful development:
-a text editor
-and program to deploy your binaries
I’m not commenting on the text editor, everybody has their own favourite one.
The toolchain can be realised with the standard cygwin environment, but my approach is a little different and inspired by the new IDE: I take the windows binaries of arm gcc (included in the new IDE), I make a custom makefile (similar to the “minimalist makefile” example), I use gnu make for windows and finally the python gendwl script to generate the dwl file in the end (all that is blazingly fast once set up).
Finally to download the binaries I use OpenTerminal (which I wrote and which still is buggy, but there will be updates).
This solutions has many advantages over the IDE: it works perfectly on linux, it is extremly customizable (ie. it can be automated, you can include c++, STL, etc.) and it can be “distributed” over many computers (ie. using a version control system you can work as a team on the sources, compile locally or remotly on a server, etc.).
To sum up this rather lengthy post in two suggestions (for jamie and the-furiously-fast-wavecom-team-that-created-a-new-IDE-in-only-nine-months):
-help the beginners by streamlining the compiling-downloading process and by making the IDE more customizable
-help the professional users by giving them a toolchain that works out of the box, and on linux, and is highly customizable
Almost a year passed… No answer to makefile question from Wavecom…
I don’t think that Wavecom ever said they were even considering Makefile support, which most likely means that they’re not.
M2M Studio doesn’t generate makefiles, but:
- it allows headless build of M2M Studio projects (i.e. without running the GUI)
- it runs under Linux since 1.1.0
- it allows additional customized build steps (before and after the main build)
So actually, we don’t plan to support makefile generation, unless some needs are not fulfilled by the current implementation (please elaborate if you have any need you can’t reach with today’s M2M Studio ).
Building with the ARM (rather than GCC) compiler?
Possible since 1.0.0: just update the settings in Window > Preferences > M2M Studio pages.
It actually detects the configuration from already defined environment variables, but you can override them from here.
Please tell how to do this!
Please refer to the M2M Studio online help, in M2M Studio User Guide > Build System > Advanced topics > Projects build in headless mode topic.