finding unused translations / strings ...

[Michael Meeks]

On Mon, 2012-06-11 at 23:10 +0100, Caolán McNamara wrote:
> > - STR_DOC_LOADING and STR_CHAIN_OK were unused. They are represented in 
> > many po files. Is there an easier way to remove them or only per regex/sed?
> 
> Stuff in .po files I think basically "take care of themselves", but
> maybe Andras can give the right answer for that. I *presume* there's a
> form of garbage collection taking place there so that translations of
> removed stuff doesn't continue forever.

	Yep - the only problem is that there is no link between the code
dropping a string usage, and leaving it around forever to be constantly
translated. It is quite possible that some small %age of our strings are
simply never used & just waste translators time.

	I just thought about this a bit ... and hacked up a very quick shell
nasty:

$ for a in `git grep String | grep \.src | grep -e '[RSH]ID_' | \
            sed 's/^.*[RSH]ID_/ID_/' | sed 's/[  {=].*$//' | sort | uniq`; \
  do echo -n "$a " ; (git grep $a | wc -l); done

	Not a perfect guide and of course it could be lots cleaner ;-) As a
finger-in-the-air guide though it's prolly reasonable.

	If the count is '2' or lower for hits - then (quite probably) the
string is either unused - or it is one of those gross cases Caolan
pointed out (which probably need identifying and strangling carefully
anyway). So eg.

ID_BEZIER_TOOLBOX	11
ID_BIB_STR_FIELDSELECTION	2
ID_BIB_STR_FRAME_TITLE	3
ID_BIB_STR_TABWIN_PREFIX	2

	ID_BIB_STR_FIELDSELECTION and ID_BIB_STR_TABWIN_PREFIX bear some
further investigation. Given that RID_BIB_STRING and RID_BIB_START are
not mentioned in arithmetic outside the .hrc file - I suspect we could
happily loose these (eg.)

	If someone is looking for cleanups - those sound like they might be a
fruitful place to start :-) are you interested Thomas - or should I file
an easy-hack ?

	HTH !

		Michael.


-- 
michael.meeks at suse.com  <><, Pseudo Engineer, itinerant idiot