flush grcat line writes

[dperny]

I used grc in a Docker container that runs go tests. I have a configuration that correctly colorizes my tests. However, the output of grc go test -v does not get written until the tests all finish, unless the container is run with a TTY. Adding flush=True to the print function in grcat causes the output to be written out line-by-line as lines become available without a TTY.

I don't know how this affects things like performance or what side effects it has. You may want to instead write a patch that gates this behavior behind a flag. I'm not sure what the correct behavior is. I'm using this fork for my stuff and it works fine for me.

[trinitronx]

I'm wondering if this is also the reason for the behavior of read prompt not printing in #82

[luisdavim]

I think using flush is a good idea.

[trinitronx]

So I ended up doing some debugging because I ran into bug #82 again today.

I think this PR is a good idea too... unfortunately it's not related to #82.

I tried testing with this change and adding a debug statement before it prints the line. Yet, even with print( ... , flush=True) grc still behaves the same as in #82 .

For example:

        try:
            print('GRC: printing line with flush', flush=True)
            print(nline, flush=True)

I tested this make target that uses the Bash builtin read -p 'Enter FOO: ' FOO, and with the following aliases:

$ type make
make is aliased to `colourify make'
$ type colourify
colourify is aliased to `/usr/local/bin/grc -es --colour=auto'

Running make foo still does not show the prompt. It does show the debug output happening:

make foo
GRC: printing line with flush
echo 'hello from foo.sh'
^CGRC: printing line with flush
Enter FOO: make: *** [provision] Interrupt: 2

So, this shouldn't hurt anything to use flush=True but I guess it doesn't fix #82 either. The only thing I can find that affects colourify make read -p "Prompt" behavior is removing the -es flags from colourify alias:

alias colourify '/usr/local/bin/grc  --colour=auto'