Emacspeak discussion list

[Jerry Sievers <gsievers19 AT comcast.net>]

Robert, thx for your work on the switf driver which I can tell is really
crisp, though this new Mac run the Python code even better than my few
years old Macbook Pro M1!

Some feedback below inline...

"Robert Melton" (via emacspeak Mailing List) <emacspeak AT emacspeak.net>
writes:

> Hey Jerry! 
>
> I hit C-x l and I get the output as expected, it speaks:
>
> "Page has 79 lines (16 + 64)"

Yep as it *should*.  On my box, only after issueing the C-e a
emacspeak-speak-message-again does anything sent via the message
function get spokenn.

It's not a case of dupes being suppressed if I were to repeat the same 
command.

> but interestingly if I 
> hit it again without taking any other action, meaning I 
> press C-x l (which speaks) then I hit C-x l again, it is 
> silent. 
>
> I never used Emacspeak prior to 59, so I always assumed 
> this behavior was intentional, to stop spamming of duplicate 
> output. That said, I never confirmed this was intentional but
> I know it happens with other things too.

It probably is intentional

> If you just put (message "hi there") in the scratch buffer and 
> highlight it and bind a key to eval-region and evaluate it over 
> and over you only hear the first one. 

Ack.  I did also try using the message function in the scratch buffer
and those didn't get rendered either, on the first attempt.

I have yet to try the log-mac wrapper to inspect if the message text is
going out to the driver or not.

I spent a great deal of time browsing the sources before getting around
the bisection.

I'm going to take a careful look at the byte compiler messages to see if
there's any interesting warnings coming up during compile of the
emacspeak-advice.el code where the regression seems to be.

OH, and I was very entertained by all of the swift compiler noise
building swiftmac but not going to worry about those and looking forward
to being able to adopt it once the message function shyness problem is
figured out :-)

I'm curious what some of the others think about this.

Cheers!

>
>
>> On Jun 11, 2024, at 00:32, Jerry Sievers (via emacspeak Mailing List) 
>> <emacspeak AT emacspeak.net> wrote:
>> 
>> Greetings Raman, et al;
>> 
>> Hopeing to be missing something simple here, but below is a patch that I
>> bisected out from between emacspeak versions 57 and58, where on my setup
>> at least, output from things like C-x l count-lines-page, and perhaps
>> lots of things that invoke the message function, don't get spoken.
>> 
>> Below marked with ## is the patch.
>> 
>> Some TLDR first is, that after a couple years of happiness running
>> emacspeak 56 on another Mac, I purchased a shiny new Macbook Air, 15
>> inch, M3, 16GB RAM.  sweet!
>> 
>> I grabbed the latest sources and built, at first using the Python
>> server, then tested Mr. RobertMeta's snappy swiftmac driver.
>> 
>> I had noticed right away that regardless of swiftmac or the Python
>> driver that $some things weren't speaking but lateer realized that
>> apparently many things, that invoke the message function didn't speak.
>> 
>> I went on to roll back to v56 OK, v57 OK using the Python driver, then
>> v58 at which point some essential things went silent.
>> 
>> I'm aware of the newer side-channel notification work but enabling that
>> for the swiftmac driver made no difference, and with the log-swiftmac
>> wrapper, confirmed that some test stuff I ran using the message function
>> never made it to the driver via regular or notification instance.
>> 
>> The bisect gave me this patch as first regression case from between v57
>> and v58.
>> 
>> ## commit 14d5e4f59
>> Author: T.V Raman <tv.raman.tv AT gmail.com>
>> Date:   Mon Mar 13 09:20:26 2023 -0700
>> 
>>    protect against echo-keystrokes being 0
>> 
>> diff --git a/lisp/emacspeak-advice.el b/lisp/emacspeak-advice.el
>> index f53e4f43d..67830d7a1 100644
>> --- a/lisp/emacspeak-advice.el
>> +++ b/lisp/emacspeak-advice.el
>> @@ -775,8 +775,10 @@ When on a close delimiter, speak matching delimiter 
>> after a small delay. "
>>               (not (zerop (length m)))
>>               (not (string= m emacspeak-last-message))
>>               (not (string-match ems--message-filter-pattern m))
>> -              (<  (/ echo-keystrokes 20)
>> -               (float-time (time-subtract (current-time) 
>> ems--lazy-msg-time))))
>> +              (and
>> +               (not (zerop echo-keystrokes))
>> +               (<  (/ echo-keystrokes 20)
>> +                   (float-time (time-subtract (current-time) 
>> ems--lazy-msg-time)))))
>>            (setq ems--lazy-msg-time (current-time)
>>                  emacspeak-last-message  m)
>> ;;; so we really need to speak it
>> 
>> 
>> Please enlighten me or advise on how I can further research this.
>> 
>> Thanks!
>> Emacspeak discussion list -- emacspeak AT emacspeak.net
>> To unsubscribe send email to:
>> emacspeak-request AT emacspeak.net with a subject of: unsubscribe
>
> --
> Robert "robertmeta" Melton
> lists AT robertmelton.com
>
> Emacspeak discussion list -- emacspeak AT emacspeak.net
> To unsubscribe send email to:
> emacspeak-request AT emacspeak.net with a subject of: unsubscribe