text-to-speech in firefox (Re: firefox-116)

[Greg Troxel <gdt%lexort.com@localhost>]

pin <voidpin%protonmail.com@localhost> writes:

I adjusted the title to describe what the message is about

> Just built firefox-116 and it brings in 9 (nine) new dependencies when compared to 115.
> In my system, these are:
>
> -dotconf-1.3
> -espeak-1.48.04nb5
> -flac-1.4.3
> -libao-1.2.2
> -libao-sun-1.2.2nb2
> -libltdl-2.4.7
> -libsndfile-1.2.0nb2
> -mpg123-1.31.3
> -speech-dispatcher-0.10.2nb7

FWIW, the set of these that were not already on my desktop built in 12
minutes (and my box is really old), compared to much longer for firefox
itself.

As far as I can tell these are all Free Software and not offensive.

> I'd guess this is related to speech-dispatcher which, according to the
> commit message is not tested yet.

I would expect that this really means "the firefox I am committing works
but I haven't tested the TTS part".  Which seems fine; it being in the
build eases others trying it.

> Shouldn't/couldn't this be a build option for those who would actually want to have TTS (text-to-speech)?
> Or, is thin enforced by Mozilla and can not be disabled?

I don't see why you are jumping to conclusions that this should be
optioned.  We generally have options when there are good reasons to have
it one way or the other which outweight the maintenance burden of
maintining conditional PLIST entries etc.   firefox is already difficult
to maintain.

Also, options don't play well with binary packages.  The default
controls what binary packages do, and there the build cost is less of a
concern.  So it is 95%+ clear that if it were an option, it would
default on.  But the way your phrased it ("actually want", which implies
that reasonable people would not want), it sounds like you think it
should be default off.

Without knowing what's going on, TTS seems like a useful feature,
especially for those with vision impairment, and if firefox in general
has this, it seems that the default pkgsrc build should too.  So if it
were an option, it would be on.

Options that default off tend to be things that are rare or odd,
relative to the cost of them being on.  That might be a qt5 GUI for
something that is basically command line, changing the package
dependency footprint from super lean to enormous.

It's a little hard to tell what the total change in dependency footprint
is, but a quick look makes me think it's fairly minor, especially if one
considers build tools.  By that I mean, do a limited pbulk with firefox
as is, and one with this optionized and off, and compare total time,
total size of packages built, and total size of packages that are
installed when you install firefox from that bulk run.


Overall my impression is that the basic issue is that firefox is not for
you :) It's large and fairly bloated, requires large and memory-piggy
tools to build, and tends to do things itself rather than rely on system
mechanisms.  I see on one side deciding to be ok with that and on the
other deciding not to run it.