LeafLabs Garden

A place to share, learn, and grow...

1. 

   mbolivar
   LeafLabs

   The temperature of the conversation is starting to get pretty high (and we're way off topic), so I'll just explain our position and hopefully we can all leave it at that (and use this thread as a pointer for when this sort of question inevitably comes up again).

   We like distributed version control systems (DVCSes), and we like Git. It suits our workflow, and we think using it makes it easier for the community to hack on libmaple, share with others, and give back, especially combined with the truly exceptional work the people at GitHub are doing.

   Practices from centralized VCSes like embedding timestamps into files just don't make any sense in a distributed system; understanding why that is true is part of coming to grips with DVCS.

   If you want to hack on libmaple, I encourage you to just fork the Git repository and track your changes in that way. This takes some getting used to, but it's well worth it. poslathian once described the excitement you get when you finally "get" Git to be like finding Excalibur, and I totally agree. Using Git also makes sharing your changes with the rest of the world (and getting them merged upstream into the "official" LeafLabs tree) much easier, which is good for you and for us.

   However, if you don't want to, you are of course welcome to use any other method you like. However, we don't have the resources to support all the world's versioning systems, so if that's the route you take, then you're on your own.

   Posted 4 years ago #
2. 

   BruceF
   Member

   @mbolivar:

   Apologies for my part in raising the temperature. On a technical basis, I agree 100% with your position and I'm glad to hear it from you. I will now revert to lurk mode.

   Posted 4 years ago #
3. 

   gbulmer
   Moderator

   samtal - you said "The PWM accuracy and stability is insufficient as an option"

   Approximately what sort of accuracy and stability are you aiming for?
   Is the STM32F103 'High-Density Family' parts' DAC good enough (dual 12-bit max 0.25Msps, I think)?
   How much lower resolution and sample rate could you live with? I ask because an external DAC might be feasible, or even some (highly dubious) mixing of DAC and PWM.

   [Side note on version control
   I used CVS and relatives in the 80's and 90's. I believe I understood why we put change logs into the source code.
   I don't think our reasons apply to git. All of the change logs are still available from git, just not in each source file (use git log, etc).

   Putting version numbers into text strings in source files was very helpful when we distributed binaries to customers. We could tell exactly version they were using by extracting the 'string's. I think there are better ways to do that.

   I did a lot of work with Product Definition, Change Management and Bill-of-Material systems for motor cars in the 90's. (Product Evolution Through Time, or Product Lifecycle Management.)

   Every car made anywhere in the world (which wants to sell to the USA, Japan or Europe) has a unique Vehicle Identification Number (VIN). That single string of letters and digits identifies that specific car. If the systems are good enough it identifies the exact version of every part, that engineering chose to track, and the exact version of every build instruction performed, and every tool used, that manufacturing choose to track. Those identifiers lead to every CAD file, and on to the engineers, suppliers etc. If the systems are really good, it could be used to identify every effective change notice, and every person who authorised those product, engineering or manufacturing change notices, and every contract for every part, which culminated in that specific car. (I wish governments did this :-)

   It is important for a car that all of that information, and those relationships are tracked.

   None of that information *HAS TO BE* stored in the source files (CAD files, part drawings, build instructions, etc. or written on the parts etc.) if the systems are good enough.

   HTH]

   Posted 4 years ago #
4. 

   samtal
   Member

   OK, guys, I surrender and capitulate.
   Let's get some work done instead.
   ( I want to keep my credit for serious questions instead....)
   samtal

   Posted 4 years ago #
5. 

   gbulmer
   Moderator

   samtal - you said "The PWM accuracy and stability is insufficient as an option"

   Approximately what sort of accuracy and stability are you aiming for?

   Is the STM32F103 'High-Density Family' parts' DAC easily good enough (dual 12-bit max 0.25Msps, I think)?
   How much lower resolution and sample rate could you live with?
   I ask because maybe an external DAC might be feasible, or even some (dubious) mixing of DAC and PWM.

   I had a look for DACs with a similar or better spec than the one in the STM32F103xC (upwards) start at $4.12 (I did the same seach at Farnell UK and found them start at £2.36):
   Newark 256ksps to 1Msps, 12bit to 16bit DAC

   The SPI interface is quite quick (18MHz clock), so it may be just about feasible.
   Is this practical for you?

   (I accept it would be lots less hassle using a STM32F103RE)

   Posted 4 years ago #

RSS feed for this topic

Reply

You must log in to post.