[PATCH] qmake: use a dedicated build script to generate ssrf-version.h

[Lubomir I. Ivanov]

On 31 January 2015 at 17:56, Dirk Hohndel <dirk at hohndel.org> wrote:
> On January 31, 2015 5:54:36 AM "Lubomir I. Ivanov" <neolit123 at gmail.com>
> wrote:
>
>> On 31 January 2015 at 15:29, Lubomir I. Ivanov <neolit123 at gmail.com>
>> wrote:
>> > On 31 January 2015 at 00:45, Cristian Ionescu-Idbohrn
>> > <cristian.ionescu-idbohrn at axis.com> wrote:
>> >> Try this same thing with adding:
>> >>
>> >> set -eu
>> >> #set -x #xtrace
>> >>
>> >> Any of these may fail in absence of
>> >>
>> >> [ $# -eq 3 ] || {
>> >>         echo "ERROR: missing argument(s)"       # alternative, usage
>> >> message
>> >>         exit 1
>> >> }
>> >>
>> >
>> > i know of this, but it was problematic because $# seems buggy under
>> > MSYS, invoking from GNU-Make $# == 2, invoking from the cmd-line $# ==
>> > 3.
>> > thus, the above error checking is non-portable and it simply fails.
>> >
>>
>> ok, i was able to get this to work, because it was too weird and the
>> number of argument is important.
>> CMD is the reason for the bad argument count, it passes an extra
>> argument for some reason and the equality check fails, but given users
>> will call the script from the MSYS shell or from a Makefile and not
>> from CMD directly this works:
>> (also -lt can be used as we ignore the rest of the arguments)
>>
>> set -eu
>> #set -x
>>
>> if [ "$#" -lt 3 ]; then
>>     echo "ERROR: missing arguments";
>>     exit 1;
>> fi
>>
>>
>> not setting #xtrace as the other scripts don't do that.
>> new patch attached.
>
>
> Thanks for working on this.
>

no, problem at all. it will be useful for my own qmake based project as well.

> All of my build scripts plus the way I package things for the various Linux
> flavors depend heavily on the current way qmake works.
>
> So I'm not going to take this patch before 4.4 and risk having to redo and
> retest all of that.
>
> I'll take it after 4.4
>

absolutely, i would do the same.

lubomir
--