Re: Stop making unneeded improvements

[Lamar Owen]

[Accidentally sent this from the wrong account.... this is the right 
one....]

On 1/5/21 10:30 AM, Glen Langston wrote:
> Concerning my previous email, sent in extreme frustration. I 
> understand that we all want to make improvements and progress. 
I want to take a moment and thank Glen for posting these messages. I had 
no idea his group was doing some of the things that they are; what a 
timely discovery, since I'm working on better backends for our radio 
telescopes as well, and what Glen has pointed to is very well written 
and implemented as far as I can tell.

I want to echo his frustrations; sometimes it seems things move too 
fast, and sometimes too slow.  I have used various versions of GNUradio 
over the years, and it's been one of the packages that should be the 
poster child for the "don't try to remember anything you learned, since 
it's all going to change anyway" school of development. Python 2 to 
Python 3 is bad enough; although I did a port to Python 3 of an antenna 
control shim for a program to interface a azimuth-elevation control 
setup to run with the right ascension-declination controls of two of our 
instruments; again, I understand the Python devs reasoning but man it 
was a pain to get all the new syntax and semantics (especially for 
string handling, from UTF to byte strings for the TCP/IP backend control 
protocol!) right and just working again; and that was just a few hundred 
lines of code.  And I know you, the GNUradio developers, had it even 
rougher with all the changes; I sympathize.

GNUradio has a large set of dependencies, for sure, and this has always 
been the thing that has made using it on various distributions rather 
difficult at times.  Having to rewrite a module because you updated the 
underlying operating system, and thus dependencies A, B, P, and Z all 
got updates that requires a GNUradio version update that requires a 
rewrite and maybe even a certification/calibration test of the new 
module.... it does get frustrating.

Updates shouldn't break things willy-nilly!  (this is also the case with 
another piece of software I use, PostgreSQL, where backend data formats 
and modules can break between major versions).  I personally would have 
liked GNUradio 3.8 to have carried a version of 4.0 to indicate that it 
is a major version upgrade; if moving from Python 2 to Python 3 wasn't 
enough to trigger that version bump, what is?

I'm trying to debug and patch the UHD stack for USRP1 with DBS-RX 
daughterboard; it just quit working, and while I've reported to the UHD 
list no action has been taken to fix the regression.  The DBS-RX has 
been calibrated by us for this application, and now I'm going to have to 
calibrate (or rather, have a couple of spring/summer interns enrich 
their learning experience by doing this calibration) another SDR; 
probably a modified ADALM-Pluto (GPS-locked clocking mod using a Leo 
Bodnar GPS-DO to drive the sample clock) or a LimeSDR; the frequency 
ranges I need are 1.2GHz to 1.7GHz (HI line spectroscopy with larger red 
and blue shifts for extragalactic source work) and 2.0GHz to 2.3GHz.  
SDRPlay, AirSpy, and most of the cheap RTL-SDRs won't tune that last 
range; DBS-RX does and worked well for us back of now.

Yes, I know.... PROGRESS.  And progress is certainly needed; but 
progress can be painful.  So while I'm not complaining per se, I do want 
to echo the concerns that especially Glen has raised.
--


pari <https://www.pari.edu>

        
Lamar Owen, /Chief Technology Officer/
Pisgah Astronomical Research Institute
/A not-for-profit public organization/
Email lowen@pari.edu <mailto:lowen@pari.edu>
Telephone 828.862.5554 <tel:828.862.5554> | Fax 828.862.5877 
<fax:828.862.5877>
1 Pari Drive, Rosman, NC 28772
www.pari.edu <https://www.pari.edu/>
fb <https://www.facebook.com/PARIastronomy/>ig 
<https://www.instagram.com/pariastronomy/>in 
<https://www.linkedin.com/company/pisgah-astronomical-research-institute/>tw 
<https://twitter.com/Astronomy_PARI>yt 
<https://www.youtube.com/channel/UCZ14Fw8C8DZAnTvk8IZ7yXg>