Hi.  I apologize for the delay and reply below:

>> And why does <S-iso-lefttab> move backward but <S-C-i> (which ought
>> to be equivalent to <S-iso-lefttab>, no?) moves forward?
>
> You'd think they'd be equivalent, but actually they are not.  You can
> use C-h k to test, even in a GUI/X11 emacs.

I am aware that ~C-h k~ differentiates between <S-C-i> and <backtab> (or
<S-iso-lefttab>), but I still think that, intuitively, since <C-i> is
equivalent to <tab>, then the user expects <S-C-i> to be equivalent to
<backtab>.  There is no use in <S-C-i> having the same behavior as
<C-i>.  And if the user is navigating forward with <C-i>, then, if she
moves past the target and thus needs to go backward, it is convenient to
press <Shift> and hit <C-i> again (and vice-versa).  And <S-C-i> moving
backwards is consistent with Org Mode tables, where <S-C-i> moves
backward but <C-M-i> is bound to `complete-symbol'.

But maybe I should not press this point further.

>> Also, it seems that <SPC> moves forward and <DEL> moves backward, but
>> <S-SPC> does nothing useful.  For consistency with Emacs Info Mode,
>> Emacs View Mode, and some other applications such as Firefox, perhaps
>> <S-SPC> should move backward too.
>
> I note that S-SPC is not accesible in terminals.  But if there's
> concensus that is is useful I don't mind adding it.

Binding S-SPC to something useful does not help on a text terminal, but
does no harm either; and it helps on the GUI, which most people use.

>> The introduction of help buffer for Notmuch-Show (reached by hitting
>> <?>) describes the <SPC> command without mentioning that it can archive
>> the thread; only in <SPC> specific description (which does not fit in
>> the same window; one has to scroll) does it mention the archiving
>> behavior.  I believe the archiving behavior should also be mentioned in
>> the introduction, lest the user unwittingly archive a thread she wasn't
>> supposed to.
>
> I personally don't think archiving is that destructive, if someone wants
> to improve this wording, it is the docstring for notmuch-show-mode

See the attached patch---which also adds another minor clarification.

>> Finally, the binding for <C-tab> in Notmuch-Show and Notmuch-Hello
>> inconveniently clobbers the tab-bar-mode binding for that key.  Should
>> not Notmuch be content with <C-M-i>/<ESC tab>/<S-iso-lefttab> (and
>> possibly <S-C-i>) and leave <C-Tab> alone?
>>
>
> Notmuch has used the binding since 2010 (almost 10 years before
> tab-bar-mode existed), so people might be used to it; on other hand it
> does seem redundant since the binding S-tab is also there.  I'm not a
> very serious user of notmuch-hello mode, and I don't use tab-bar-mode at
> all, so I'd welcome feedback from other people about how attached they
> are to C-tab.

In my opinion, vanilla Emacs keybindings (even if recent) should not be
overridden without strong reason.

Feel free to give feedback on whether I am nitpicking, especially if the
feedback is reasonably detailed.

Regards!

-- 
- Many people hate injustice but few check the facts.  This provokes
  misinformation.  Ask me about <https://stallmansupport.org>
- Please adopt free/libre formats like PDF, Org, LaTeX, ODF, Opus, WebM and 7z.
- Libre apps for AOSP (Replicant, LineageOS, etc.) and Android: F-Droid
- https://www.gnu.org/philosophy/free-sw.html "What is free software?"