On 2017-05-02 17:53, Roman Yeryomin wrote: > On 2 May 2017 at 10:25, Felix Fietkau <n...@nbd.name> wrote: >> board_name already exists, and I've already converted e.g. the lantiq >> target to use it directly from functions.sh. >> board_model is not used, neither is the model function from any of the >> messy copy&paste target files. Which means you're arguing for having a >> separate source file for a function that in its current form only has >> one single line of code. >> >> In my opinion the best course of action is to nuke your patches 1-3 >> completely, rework patch 4 to adjust the generic code as proposed, and >> adjust patch 5 to simply use the board_name() from /lib/functions.sh >> >> Just ignore the board_model nonsense, it does not need to be part of the >> shell 'API' at all. > > I'm talking about board_name, not model. I've added model to the file > only because /tmp/sysinfo/model exists. > You say that including 365 (non-functional) lines is better than 15 > (functional) when you only need board_name? I cannot understand > that... I'd say almost all places that will use board_name() already need /lib/functions.sh. /lib/functions/uci-defaults.sh already includes it, so do many other core include files. Among the files touched by your rework, I think there is only one file that does not already include functions.sh.
In functions.sh we already do have quite a few functions that are used only in some places and not in others. Putting every small piece into a separate include file would make things messy very quickly. This may seem like a minor detail to you, but I want to make sure that we don't add more unnecessary clutter here. - Felix _______________________________________________ Lede-dev mailing list Lede-dev@lists.infradead.org http://lists.infradead.org/mailman/listinfo/lede-dev
