Re: FDA Approval
|
I don't think anyone is arguing for one source code file. Marcelo has divided them into files where the primary file name suggests what the file contains. Some are less clear (hal.?) and a comment a the top of each file saying what the reader is about to read would be helpful. Also, some comments in the code would be helpful, too. I have argued elsewhere for years about NOT using a single INO file. One big reason is because the IDE supports incremental compiles. Recently, I worked on a project with 19 source files and the first compile in the morning on a pretty fast machine would take almost a minute. When debugging, you tend to recompile the same file over and over. That file then has a "dirty flag" set by the IDE, while the other 18 files didn't. The GCC compiler is smart enough to cache the other 18 files and just compile the dirty file and then relink. Often, recompiles take only a couple of seconds because of this. Given enough compiles and enough days, the time savings is huge. All I'm saying is: When you name a file, think about a name that suggests the content of the file, place a comment at the top of the file with a short description of what the file is for, and add some comments where it would help explain what the code does. Jack, W8TEE
On Friday, April 17, 2020, 3:37:13 AM EDT, Zeeshan Abbasi <zeeshan@...> wrote:
Whilst totally ignoring the personal talk as that is not my concern, I will just add a bit about the verbosity of the code. The code has a lot of files and that is making it quite modular. So those with different hardware and sensors can change one specific file and they are done. One large file is a maintenance headache when 5 people are sending pull request of various features it can have deadly consequences if a merge goes wrong. In fact once stable and fully running we also will need things like test coverage and few other bits so that others can see how to modify and utilize parts of the code to their own setup and be sure that nothing is broken. I am regularly merging from the main with my own setup that had different sensors but now as I am finding the right ones and getting closer to the main (still differences). The modularity is decreasing dependency and each pull usually only creates a couple of small merge conflicts. One single file would have a lot of code movement in each merge resulting in a lot of waste of time and risk of a missing bit. I think this is a critical system and a couple of days to understand the code (for those that need assistance) is an investment worth making. Also what is this group for if they cannot seek help and ask questions. Also the loop being very fast and non-waiting means the UI and navigation stays responsive. ? From: [email protected] <[email protected]>
On Behalf Of jjpurdum via groups.io
Sent: Friday, 17 April 2020 7:08 am To: [email protected] Subject: Re: [VentilatorDevelopers] FDA Approval ? Actually, you're not sorry at all, as evidenced by your tone in the rest of your email. I'm not sure who wrote the code for holding a button, but that's of little matter and I'm pretty sure it wasn't done to kill the patient. My guess is it was done to get around some kind of resource limitation, but I don't know for sure. And, sometimes code is designed to do nothing for milliseconds, so I doubt that a blanket statement against it is warranted. As to contacting you, you made it very clear you were unapproachable by constantly reminding us how busy you were and had more important work to do for your employer. I think we both share the goal about saving lives, we may just differ on how that might actually happen in the field. To that end, you might start adding some comments to your code so that every one else in the world who is less talented had some hope of making it work if they, for whatever reason, are forced to work with different hardware in a vastly different environment. ? As to the friends comment, it sounds like you could use one...let me know if that's the case. ? Jack, W8TEE ? On Thursday, April 16, 2020, 4:22:15 PM EDT, Marcelo Varanda via groups.io <mv_email@...> wrote: ? ? Sorry if you feel like that. I would be glad to collaborate to any code that could be used in production. I have done this in many open source projects. A code that you press and hold a button and kill the patient in my opinion is a bad code. A code that does nothing for milliseconds is a bad code. You guys could have collaborated with this baseline code; making questions in case you do not understand something; even calling me; but elected to collaborate with a code that you could easily understand as written by a IT teacher in high school. I respected your choice. As per the message you just criticized I answer a question about profiles. I am more interested in good and safe results to save lives. Making friends comes in second place. |
jjpurdum