Re: [OT] ESA project memories (was Re: is Ada used in James Webb Space Telescope software?)

[John McCabe <john@nospam.mccabe.org.uk>]

On Fri, 31 Dec 2021 23:18:49 +0200, Niklas Holsti wrote:

> On 2021-12-31 12:26, John McCabe wrote:

<snip>

>> We didn't really have that problem. On TCIU most of our requirements
>> came from Dornier - > MMS-UK (ASAR instrument prime) - > Alcatel - >
>> MMS-UK (TCIU team). Both MMS-UK teams were in Portsmouth.

> Interesting :-).  I had a similar, but inverse, experience in a later
> project (SW for the Flexible Combined Imager instruments on the MTG
> satellites) where Thales Alenia Space (France) was both our customer for
> the whole SW and our subcontractor for a part of the SW. It led to a
> number of "direct" communications and decisions between the two TAS-F
> teams that bypassed our team (in Finland) and of which we learned later.
> But not much harm done, overall a good project.

It would be inappropriate of me to say whether or not that sort of 
behaviour occurred on ASAR, although I seem to remember occasions where 
Alcatel waived their right to be piggy-in-the-middle as some of the 
discussion about SAR pulse timing and the effect of shifting things 
around a bit, to deal with the fact that we would've needed a mid-90s 
supercomputer (and a substantial re-design of the TCIU -> T/R Module 
interface) to achieve what was originally specified, would've fried the 
brains of the people who were actually involved :-)

<snip>
 
> Although splitting work up into several companies does easily make for
> inefficiency, in can also have the benefit of documenting stuff that
> otherwise might be lost in internal e-mails or face-to-face discussions.
> That is, if the companies involved do their work properly, and don't act
> as you describe for Alcatel. But perhaps the Alcatel technical people
> did as well as they could to mitigate a poor higher-level decision, by
> being basically a transparent conduit, as you describe.

To be fair (to MMS!), the actual documentation that was produced at the 
instrument level was pretty good. To be fair to Alcatel, as I mentioned, 
we'd been working without them on this for a long time before ESA decided 
to mandate that they should "manage" the TCIU development as a 
subcontract, so they were forced to pick up on stuff they pretty much 
hadn't cared about before.

Ironically none of this helped with the documentation from Alcatel; the 
TCIU -> T/R Module interface I mentioned, for example. We went through 3 
rounds of TCIU Software Requirements reviews (i.e. SRR, then re-visited 
at ADR and DDR or something like that), where our assumptions on how that 
interface worked (based on rough sketch ideas we'd been given rather than 
formal specification) were described, before someone at Alcatel bothered 
to read it and say "nah, doesn't work like that" (presumably in French) :-
)