Hi guys,
I hope I won't hurt anyone with this (very long, sry) message but I am so confused among the various free and non free libraries about Cortex-M on the market, I need some clarification from inner professionnals.
Well, part of my work is about to code low/high functionnal features on STM32, more specifically on L0/L1 and F4 Cortex-M (totally different purposes).
I use so far the Standard Library provided by ST .
I have already noticed that parts of their library are sometimes buggy/undocumented/uncompleted but there is a lot of ressources and example which help to integrate functionnalities in a faster way.
It recently seems, as everybody knows, that ST tends to provide better and less buggy implementation of their high level library through CubeMx and better bridge/integration of 3rd part library. Hi guys,
I hope I won't hurt anyone with this message but I am so confused among the various free and non free libraries about Cortex-M on the market, I need some clarification from inner professionnals.
It recently seems, as everybody knows, that ST tends to provide better and less buggy implementation of their high level library through CubeMx and a better bridge/integration of 3rd part library.
Specifically regarding the USB library, implementation of FATFS & FreeRTOS and the CPAL I2C StdLib.
Comparing the offer from Keil pro-licencing vs ST StdLib, I'm asking myself the true benefit of getting the first one.
I Give a personnal study case which I think is relevant for the described situation: I recently developped an application on L1 using USB CDC & I2C, the entire code is based on ST StdLibrary. For cost & space purposes, I want to try this code on the newly commercialized L0.
I2C & USB library are not available through standard dwnl but these functionnalities could be generated within CubeMx.
So I try to get my code working and... I realize that CubeMx high level layer API at least for L0 are not mature at all (lack of example, terribly buggy, sometimes downgraded compared to the older StdLib,...).
Thus in order to get my code working properly I have 2 choices: 1. Debugging the ST code on myself (challenging but unfortunatly no time for that), 2. Getting a commercial High API level like Keil Pro, which I hope provide USB CDC for L0;
Regarding the efforts and the cost I have to put (especially the effort, the cost belonging to shareholders...), and admitting that in 6 to 12 months the CubeMx will be mature specifically regarding the USB library, implementation of FATFS & FreeRTOS and the CPAL I2C.
Giving a personnal study case: I recently developped an application on L1 using USB CDC & I2C, the entire code is based on ST StdLibrary. For cost & space purposes, I want to try this code on the newly commercialized L0.
I2C & USB library are not available throught standard dwnl but these functionnality could be generated within CubeMx.
So I try to get my code working and... I realize that CubeMx high level layer API at least for L0 are not mature mature at all (lack of example, terribly buggy, sometimes downgraded compared to the older StdLib,...).
Then in order to get my code work properly I have 2 choices: 1. Debug the ST code on myself (challenging but unfortunatly no time for that, 2. Get a commercial High API level like Keil Pro, which I hope provide USB CDC for L0;
Regarding the efforts and the cost I have to put (especially the effort, the cost belonging to shareholders...), and admitting that in 6 to 12 months the CubeMx will be mature, is it worthy to buy the pro-license to have the high api layers working properly (I hope ;)) untill ST will release a proper high layer API ?
In a more global point of view, (I still don't want to hurt anyone), ST is releasing more and more high level API for their STM32 which could override the others proprietary API (Btw I noticed that in the F2 and F4 Keil device pack, the ST StdLib are proposed), so what will be the advantage of using commercial API rather than free ST Libs when only using ST chips? (better support ? Less buggy ?)
It ends by an opened question :)
Thanks for reading and further answering this post.
Regards.
Most of these high level abstractions, and code generators are aimed at people who don't really want to understand how the underlying hardware actually works, or have the experience to know any better. The sales pitch is they make it all easy, the reality is you're still going to have to get your hands dirty and fix things, and the engineer at the tool vendor is unlikely to have months to understand the nuances of a 100 different micros, or configuration combinations thereof.
The cost of the tools is generally fractional compared to the engineering/company resources to bring a product to market. They are less apparent because they are buried.
I'd wager that the Keil software developers have got far more embedded coding experience than the ST devs working on CubeMX. You pay your money, you make your choice.
I see I must have activated some Keil censorship operations when I did write a post about some chip application engineers not being professional software developers.