Re: SDR branches selector

[Marcus Müller]

Hi Glen,

you know how much I like that you're building great software for science and 
education
based on GNU Radio, and it's exactly why (I back then, but now) we didn't let 
3.7 just die.

So, I think, we really do what you wish for: GNU Radio 3.7 is still around.

Of course, old software ages; it's literally technically impossible to fix a 
lot of the
stuff in 3.7, so we must advise against starting to use it.

We, from an organizational point of view, also can't add new features to 3.7, 
either,
because that requires backporting software to languages that aren't supported 
anymore,
i.e. every developer would need to write blocks twice. Together with the 
compatibility
effects of that, this more than doubles the development effort. Worse even, 
effort is at
least polynomial relative to the number of different releases you have, as 
there's
combinations of versions and platforms.

Fixing something in a way that fundamentally breaks your 3.7 flowgraphs, but 
still calling
it 3.7 would be worse than fixing it in a currently supported version. People 
rely on 3.7
to be stable.

So, the fact that 3.7 is not what I recommend beginners learn is really because 
I love
beginners to stay with the project and do SDR, instead of trying to install an 
old
environment on their new machines, and because I know what we could only fix on 
3.8 and
3.9, but not 3.7, and it makes a difference. In this particular example, 3.8 
changed the
"selector" block. Couldn't do that on 3.7, unless we wanted software using a 
previous
version of the block to just stop working.

I know, and I appreciate, that you're seeing this from "it is a lot of effort 
and it works
on my system, so please don't break it", but we also have to see it from "it 
cannot work
on current out-of-the-box Linuxes, no matter how much effort we'd put into it", 
and that's
why we *have* to move forward.

I hope you understand that this is a trade-off nobody did lightly, but 3.7 is 
stuck,
literally. It can't move in any direction; you're asking us to still make great 
additions
- but that was and is simply impossible, as much as it hurts us as people who 
work on
things, just like you, and as much as it hurts you and our other users. I wish 
it wasn't
that way, but you assuming we could just continuing working on 3.7 and making 
it work
forever is sadly not helping this fact.

Best regards,
Marcus


On 06.07.21 15:56, Glen Langston wrote:
> 
> 
>> On Jul 6, 2021, at 7:37 AM, Marcus Müller <mueller@kit.edu> wrote:
>>
>> Hi Yasir,
>>
>> On 06.07.21 06:07, Yasir ABBAS wrote:
>>> Hello community
>>>
>>> I am using Gnuradio 3.7.
>>
>> Please don't. GNU Radio 3.7 is our legacy release, which we only still 
>> maintain because
>> there's existing systems based on it.
>>
>> If you're now learning GNU Radio, you should be using GNU Radio 3.9 or 3.8, 
>> never 3.7.
>>
> 
> Thanks for all your efforts Marcus (and all),
> 
> However I still feel that the GNU Radio project made a major error breaking so
> much code in the transition from 3.7 to 3.8 and 3.9.  While we can each 
> struggle
> to update our own packages, it is very difficult to fix other’s packages that 
> we
> depend on.
> 
> Please, Please, Please, in the future, make great additions, but 
> without destroying older code.  This may mean that you’ll have multiple
> methods for new system functions, but my guess it is much easer for ace 
> developers to
> deal with parallel systems than application specific programers/developers.
> 
> Thanks
> 
> Glen
>>
>> On GNU Radio 3.7, there isn't. On 3.8 and 3.9, we've got one.
>>
>> Best regards,
>> Marcus
>

smime.p7s
Description: S/MIME Cryptographic Signature