On Thu, 7 Jan 2016 13:41:58 +0100 Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
On 05.01.2016 07:29, Stefan Tauner wrote:
On Tue, 5 Jan 2016 01:54:52 +0100 Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
The alternative would be to provide an easy way for the user to build sets of programmers supported by the respective libraries, e.g. make CONFIG_LIBUSB0 CONFIG_LIBUSB1 would build all programmers requiring libusb-0.1 and libusb-1.0 (only). We should of course give the user a good indication of what libraries would actually work but if he does not want to or can not install all of them he should (still) have an easy way to disable programmers that do require certain libraries (even without auto-detection).
The most important part (and the one I would like to focus on for now) is to provide better diagnostics in case not all libraries are available but the user actually wants them to install/fix.
Definitely.
Signed-off-by: Stefan Tauner stefan.tauner@alumni.tuwien.ac.at
Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
I have found some minor additional bugs and committed it after fixing those in r1911, thanks
+debug_shell = $(shell export LC_ALL=C ; { echo 'exec: export LC_ALL=C ; { $(1) ; }' >&2; { $(1) ; } | tee -a $(BUILD_DETAILS_FILE) ; echo >&2 ; } 2>>$(BUILD_DETAILS_FILE))
I'd love to learn how this works. Do you have any pointers?
It gets a command line in form of a string as first parameter, notes that it will execute that command by echoing it to stderr. It then goes forward and executes the command copying its stdout to the build details file. It then adds a line break for better formatting to stderr... and all that stderr output (the echo in the beginning as well as all stderr from the execution of the command) will be appended to the build details file as well. I definitely should have commented that "better" :)
Well, now it's at least present in patchwork.
Will comment on it in a future patch as well.
BTW why don't we set LC_ALL=C globally? Please don't tell me because of i10n of compiler errors...
Mostly because I didn't have that idea. Such a change would have my immediate Ack (provided it works).
I guess it works... http://lkml.iu.edu/hypermail/linux/kernel/0912.3/00246.html