On 8/7/20 1:51 AM, Erik Kaashoek wrote:
On Thu, Aug 6, 2020 at 03:21 PM, Roger Need wrote:
From what I read some users just wanted edy555's code licensed using GPLv3 to
be released in accordance with the terms of the license which says modified
source code is to be included with ANY firmware release that is made available
to others. The license terms are clear and it is not fair to the original
author (edy555), other contributors or in the spirit of "open software" to do
otherwise. It would not have been that difficult to put the modified source in
with their firmware in a zip or rar file. That would have met the license
terms and avoided all the controversy
As someone that is actively maintaining various GPLv3 SW packages I'd like to add some words.
The issue seems to be foremost a matter of tone.
Nobody here will disagree that GPLv3 SW comes with certain obligations. But at the same time someone contributing may feel uncertain about how in a transitional state the code may look and does not want to publish every time a small fix has been made because that person wants to publish well written, well documented source. So sometimes the development (and testing) goes through a number of iterations before the outcome is deemed stable and the code is published. Legally this may not be allowed but nobody wants the alternative, e.g. no wider testing of beta releases and taking someone to court on this is unrealistic.
Now what happens when suddenly in the midst of such a high energy updating sequence someone starts to DEMAND every iteration MUST have published source. It has happened to me. I did not like the tone, it felt like an insult given all the effort I was putting into this and I did considered to delete everything and stop. People that contribute to the community are motivated by the recognition they get. Someone suddenly DEMANDING something is a very strong demotivator.
All people in this group that where releasing new FW's did publish (at least as soon as the source became stable, sometimes more often) code to github. Nobody ever stated they did not intend to do this.
So I'd kindly ask people not to DEMAND something but, while voicing respect for the huge contribution, kindly request when the next update of the github repository can be expected.
I agree...
Most software development of this type is episodic and goes in bursts. It's perfectly reasonable (if not precisely according to the letter of the law, regulation, contract) to have things in a state of flux due to debugging and push out the artifacts towards the end of the burst.
But this points up a defect in the GPL vis a vis modern software development process.
A collaborative build/test environment benefits from the use of (semi) public repositories for intermediate builds - especially for the alpha and beta testers, who are interested in *using* but not modifying the software. Perhaps the progenitors of GPL (FSF?) should contemplate this model of operation.I fully recognize that some of the leading advocates of free software (both as in beer and speech) aren't necessarily interested in being reasonable - they're trying to push a cultural change.
Not everyone runs a "commit to the repo before every test build/compile" type development operation for a variety of reasons.
And even if you did, you'd wind up with people having trouble with the build because of environment differences - which libraries do you have, which version of glibc, etc.
This is true even in professional environments. I spent a week getting NEC4.2 to compile on a Mac, even with decent instructions from someone who had done it before. Ditto when moving the code from one supercomputer to another (because the underlying environments were *slightly* different).
I have also spent weeks getting RTEMS ( a great open source RTOS) working on a new development machine - so open source is no panacea there either.
I am *quite sure* that DiSlord and OneOfEleven have their hands full just making the firmware work and aren't interested in supporting people trying to get the cross tools running, etc. Getting NanoVNA-saver working, which is in Python, which is pretty good for cross platform support, took some dozen or so messages back and forth among forum members.
--
And I also agree that there are people who demand stuff: I've experienced it, told them I'm not going to, and got flamed.
I suspect that if you DID have a build process that pushed everything to a repo before builds, but you used, say, cvs, or sourcesafe, then people would gripe about you not using git. (or, in the case of SourceSafe, gripe about using a non GPL tool to develop GPL software, which is somehow anathema by their beliefs).
In any group that is sufficiently large, there are certain to be jerks.
