[PATCH] tools: Add PKG_VERSION to sstrip

[Paul Spooren]

On 22.07.20 12:49, mail at adrianschmutzler.de wrote:
> Hi Paul,
>
>> -----Original Message-----
>> From: openwrt-devel [mailto:openwrt-devel-bounces at lists.openwrt.org]
>> On Behalf Of Paul Spooren
>> Sent: Donnerstag, 23. Juli 2020 00:15
>> To: openwrt-devel at lists.openwrt.org
>> Cc: Paul Spooren <mail at aparcar.org>
>> Subject: [PATCH] tools: Add PKG_VERSION to sstrip
>>
>> Comparing the in tree stored source file of sstrip suggests it's version 2.0[0],
>> reflect that in the Makefile.
> note that conceptually, PKG_VERSION is for _external_ packages:
>
> https://openwrt.org/docs/guide-developer/packages#buildpackage_variables
>
> "PKG_VERSION - The upstream version number that we're downloading"
>
> So, effectively this is to be used when there is some PKG_SOURCE_URL in the file.
>
> For packages that just consist of "local" code, one should just use "PKG_RELEASE".
>
> Actually, I've only recently enforced that for the package directory:
>
> https://github.com/openwrt/openwrt/commit/9c170cb92f4fbb316592c11567a080eb3f6a1fc3
>
> I'd be happy if we could (continue to) follow that same scheme for the packages in tools as well.
Works for me, I missed that specific patch but will just send a v2.
> For your ultimate goal, it shouldn't matter anyway, just replace the PKG_VERSION by PKG_RELEASE where we are using local code/there is no external code pulled.
True, either is fine.
> This obviously applies to all the similar patches you sent in parallel.
> Despite, note that the common practice for PKG_RELEASE is to use plain integer numbers, so no major.minor. I don't think that's as important as my first point, but I'd say only deviate from the plain integer numbers when having a reason (since, actually, if the code remains untouched for several years, nobody will maintain a reasonable major/minor versioning anyway, and it's easier to just bump by "1" on each change).
>
> Personally, to be honest, I'd just add PKG_RELEASE := 1 to all of the previously unversioned packages.
Isn't it cleaner to state the upstream version of a package in the 
Makefile even if the code is stored locally? I know sstrip is locally 
patch-hacked but I think it's better to state version 2.0 instead of 
release 1. Let's say version 3.1a brings some legit binary size 
improvements, no one could tell by reading release 1.
>> Motivation is the tracking of changes in the buildsystem, which requires
>> versioning of packages.
>>
>> [0]:
>> https://github.com/BR903/ELFkickers/commit/df4426a0f0ada861064d75c08c
>> bebaac7c16b3ae#diff-d3ba694d91432a068d5d3b36abf8cd0f
>>
>> Signed-off-by: Paul Spooren <mail at aparcar.org>
>> ---
>>   tools/sstrip/Makefile | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/tools/sstrip/Makefile b/tools/sstrip/Makefile index
>> 180bd1743e..99be063f4c 100644
>> --- a/tools/sstrip/Makefile
>> +++ b/tools/sstrip/Makefile
>> @@ -7,6 +7,7 @@
>>   include $(TOPDIR)/rules.mk
>>
>>   PKG_NAME:=sstrip
>> +PKG_VERSION:=2.0
>>
>>   include $(INCLUDE_DIR)/host-build.mk
>>
>> --
>> 2.25.1
>>
>>
>> _______________________________________________
>> openwrt-devel mailing list
>> openwrt-devel at lists.openwrt.org
>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel