Native FBInk: Another small tool to print crap on your screen!

[NiLuJe]

@jackie_w:

I may have skimmed the code a wee bit too fast...

* Yeah, it only writes the state to the config if retainStateOverRestart is enabled in the [state] section... :/.

* I may have totally missed how the parser handled ini sections... .

i.e., the test should probably be grep -q "invertActive = yes", since state is the section, so, a [state] header, and not a state: prefix like it's handled in the code.

----

Random comment: since here, both -C & -B are passed, one could also simply invert the arguments instead of using -h . (Not that there's anything wrong with -h: I checked ).

[jackie_w]

Originally Posted by NiLuJe

Random comment: since here, both -C & -B are passed, one could also simply invert the arguments instead of using -h .

A-ha, I'm one step ahead of you there, I just didn't admit it . I did remove the -h option and try this. Same result, or should I say, No Result

I'll try the other suggestions.

[PoP]

Originally Posted by NiLuJe

@PoP: Updated the package in #93 .

Thanks for adding the static library and header.

Originally Posted by NiLuJe

@PoP: The static lib? Wouldn't a shared one be more useful?

Absolutely. I now understand why. My FT only came with a shared library, so I can't compile with -static anyway.

[NiLuJe]

@PoP: I ended up bundling the static one because that's how the Kindle version is built, so, more practical for me ;p.

-lfbink *will* pickup the static library if that's the only one there (it prefers shared libs, but that's it). Conversely, -l:libfbink.a will *only* pickup the static library, even if the shared one is present in the search paths .

[PoP]

@NiLuJe, I understand that

but a shared compile expectedly gives

[root@kindle tcc]# tcc -Iinclude -Iinclude/freetype2/freetype -Iinclude/freetype2 -Iinclude/fbink -L/mnt/us/linkfonts/lib/autohint -lfreetype -L/mnt/us/usbnet/bin -lfbink -o  fbinkpantspatrol fbinkpantspatrol.c
tcc: undefined symbol 'fbink_init'
tcc: undefined symbol 'fbink_print_raw_data'

and a static compile seems to now need a static freetype lib

[root@kindle tcc]# tcc -Iinclude -Iinclude/freetype2/freetype -Iinclude/freetype2 -Iinclude/fbink -L/mnt/us/linkfonts/lib/autohint -lfreetype -L/mnt/us/usbnet/bin -lfbink -static -o  fbinkpantspatrol fbinkpantspatrol.c
tcc: cannot find -lfreetype

I don't want to hijack the thread with C compilation noobe stuff. Lemme learn more. Go in peace for now.

[NiLuJe]

@PoP:

1. There's a very strong possibility poor TCC doesn't know how to deal with LTO objects in the static library .

2. Never pass -static to anything, anywhere, ever ^^. -static ranges from evil to actively broken when using the glibc. It's only useful in very, very, very specific circumstances (generally, when you're static linking to an alternate libc that's designed for that, f.g., musl).

[PoP]

Originally Posted by NiLuJe

@PoP:

1. There's a very strong possibility poor TCC doesn't know how to deal with LTO objects in the static library .

Hahaha. Back to square one I lazilly thought TCC would spare me from installing the tool chain and cross compiling.

2. Never pass -static to anything, anywhere, ever ^^. -static ranges from evil to actively broken when using the glibc. It's only useful in very, very, very specific circumstances (generally, when you're static linking to an alternate libc that's designed for that, i.e., musl).

Muah-ah-ah. Noted.

Thanks

[NiLuJe]

@PoP: Switched to a shared lib.

In any case, it's smaller, so, eh, I'll consider that a win-win .

(You'll need some rpath shenanigans when compiling, or some LD_LIBRARY_PATH shenanigans when running, to properly pickup the libs at runtime ).

EDIT: And on the off-chance a recent build of TCC helps, here goes.

[jackie_w]

Originally Posted by NiLuJe

* Yeah, it only writes the state to the config if retainStateOverRestart is enabled in the [state] section... :/.

* I may have totally missed how the parser handled ini sections... .
i.e., the test should probably be grep -q "invertActive = yes", since state is the section, so, a [state] header, and not a state: prefix like it's handled in the code.

I think we're in business! The above 2 things are the key.

However, when NightMode rewrites nightmode.ini, the format of the line changes slightly.
Original .ini:

[state]
invertActive = no		# yes / no

Re-written .ini:

[state]
invertActive                   = no

So I think a regex will fix that. After a bit of googling I came up with this. Does it look OK?

if grep -Eq "invertActive\s*=\s*yes" "/mnt/onboard/.kobo/nightmode.ini" ; then

[NiLuJe]

I'll have to double-check what the busybox version of grep bundled on the Kobo supports, exactly, and I'll get back to you (dinner time), but, yeah, the idea'll be to ignore the whitespace, thanks for the config excerpt, that's what I was missing ^^.

[NiLuJe]

@jackie_w: Yep, that works!

(I always forgot what the backslash character classes refer to, so I tend to go with [[:blank:]] instead, but, yeah, \s matches a whitespace, so, that does the trick ).

[jackie_w]

Originally Posted by NiLuJe

@jackie_w: Yep, that works!

(I always forgot what the backslash character classes refer to, so I tend to go with [[:blank:]] instead, but, yeah, \s matches a whitespace, so, that does the trick ).

Thanks very much for all your help. Sorry (but not very) for being a persistent PITA As penance I'll document what I did in the main Miniclock thread in case anyone else is interested ... but maybe not till tomorrow.

Maybe I didn't need the grep -Eq because your original grep -q seems to work fine. I didn't realise that grep understood regex patterns by default.

[NiLuJe]

@jackie_w: There's a quirk involved in that it understands a specific syntax/subset of regex by default, while -E enables another syntax with more features .

And no, I can never remember which is which, so, I tend to bring out the big guns by default ;p.

[jackie_w]

Originally Posted by NiLuJe

@jackie_w: There's a quirk involved in that it understands a specific syntax/subset of regex by default, while -E enables another syntax with more features .

And no, I can never remember which is which, so, I tend to bring out the big guns by default ;p.

I think \s* is about as basic as regex gets so I'll document it with the simpler -q option.

[NiLuJe]

Released v1.8.0!

A fairly juicy release, since it introduces OpenType/TrueType rendering, thanks to @sherman's work!

It also fixes a stupid bug that broke everything on the black K4, as well as implementing the fgless mode discussed earlier .