[Replicant] Decision to not create a Replicant 6.0 SDK

Wolfgang Wiedmeyer wreg at wiedmeyer.de
Thu Apr 13 19:21:35 UTC 2017


Hans-Christoph Steiner writes:

> Thanks Wolfgang for pushing this forward!  I'm a big fan of merging
> efforts and working on a common set of Android SDK packages.  We have
> put a ton of work getting to the point where we everything is built from
> DFSG-free source.  Most of this work will directly apply to other
> distros, but some will not.
>
> There are a couple of spots that would be great to have some
> cross-distro collaboration on.  For example, an ant build for the core
> of gradle.  Right now, gradle builds gradle, which makes it very hard to
> get started, among other issues.  It should be possible to make a basic
> ant build of the core of gradle which builds just enough to then build
> gradle with.

I learned this the hard way when I tried to backport gradle for my own
use when it was only available in unstable. How did you then do the
upload to backports? AFAIR there were build errors when I tried to build
it with the version in Jessie. Are these circular dependencies the biggest
challenges or maybe the only really difficult parts for uploading to
Debian based distros? Are circular dependencies among the Android
packages documented somewhere, also the approach you chose for uploading
these packages to Debian?
We could include the ant build of core gradle as an example for this in
the post.

> Java people like to do this kind of crap. Lombok also depends on itself
> for the build, we took a similar approach there.

Yeah, I came across similar cases although I don't remember specific
ones and I was wondering if there was a trait to it.

Best regards,
Wolfgang

> .hc
>
> Wolfgang Wiedmeyer:
>> Hi all,
>> 
>> some of you might already have noticed that I was spouting my opinion
>> about how it's not worth working on a Replicant 6.0 SDK over the last
>> months[1][2]. But we never had a final verdict among Paul, Denis and me
>> or the community as a whole about this. I think we should decide if it
>> should still be a priority to create a SDK or if it should be something
>> that someone can create if they feel to, but it's not a goal that is
>> actively pursued by the project.
>> 
>> I will summarize my reasoning again below why I don't think that it's
>> worth it to work on this. I just like to have the decision on this
>> before the Replicant 6.0 release, so this can be addressed in the
>> release blog post. It might even warrant its own blog post and I would
>> prefer to do a dedicated post about it if we agree on my proposal.
>> 
>> The Android SDK is packaged in Debian, even the same API as the
>> Replicant 6.0 SDK would have. Debian even has build tools packaged that
>> are not available in the SDK, like gradle or the gradle plugin. Some
>> advantages of the Debian Android SDK over a Replicant 6.0 SDK:
>> * properly build from source without relying on Google's prebuilt
>>   binaries
>> * build is reproducible
>> * only an "apt-get install android-sdk" away
>> * Debian developers actively work on packaging more target APIs,
>>   something we basically can't do
>> * Debian Android SDK is exposed to a lot more users and has more
>>   developers, so there will be less bugs and they are fixed a lot faster
>> * Debian developers have packaged or are packaging more tools and
>>   libraries that would not be available in a Replicant 6.0 SDK
>> 
>> So my proposition is to not work on a Replicant 6.0 SDK, but to get the
>> Debian Android SDK into more distributions, especially GNU
>> FSDG-compliant ones. A blog post could address this and encourage users,
>> that are interested in a Replicant 6.0 SDK, to work with the developers
>> of the distro they are using to get the Debian Android SDK packaged in
>> their distro. If we agree on this strategy, it would be good to clarify
>> some of the things below before writing a blog post, so we can provide
>> more specific instructions on how this goal can be reached.
>> 
>> Is it correct that Trisquel and Parabola are the most popular
>> FSDG-compliant distros? It looks like the next release of Trisquel (8.0
>> Flidas) will be based on Ubuntu 16.04. Ubuntu 16.04 has the android-sdk
>> packaged[3], but it's not complete as some packages didn't make it in
>> time. android-sdk-platform-23 which is the actual target platform is
>> only in Zesty (17.04)[4]. Are there some Trisquel folks reading along
>> that could chime in on this? Would it be possible to get the rest of
>> the needed Android packages included in Flidas by backporting them from
>> newer Ubuntu releases? Are there other procedures to get packages
>> uploaded and included in a Trisquel release? The SDK will likely end up
>> in Trisquel at some point in the future, but it will probably take a
>> very long time without doing something about it now.
>> 
>> Hans-Christoph Steiner from the Debian Android Tools team addresses
>> uploading to other distros in a mail[5]. I didn't yet see any efforts to
>> get the SDK properly packaged in Arch Linux or in Parabola for that
>> matter. Arch only seems to have packages that download the SDK from
>> Google's servers without building it from source. Maybe some Parabola or
>> Arch folks can comment on this? What would be the best strategy to get
>> the packages into Arch? I guess if they are in Arch, they will show up
>> in Parabola as well, at least eventually.
>> 
>> Best regards,
>> Wolfgang
>> 
>> [1]  http://lists.osuosl.org/pipermail/replicant/Week-of-Mon-20170320/001247.html
>> 
>> [2]  https://lists.alioth.debian.org/pipermail/android-tools-devel/2016q4/002316.html
>> 
>> [3]  http://packages.ubuntu.com/xenial/android-sdk
>> 
>> [4]  http://packages.ubuntu.com/zesty/libandroid-23-java
>> 
>> [5]  https://lists.alioth.debian.org/pipermail/android-tools-devel/2016q4/002325.html
>> 


-- 
Website: https://fossencdi.org
OpenPGP: 0F30 D1A0 2F73 F70A 6FEE  048E 5816 A24C 1075 7FC4
Key download: https://wiedmeyer.de/keys/ww.asc
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://lists.osuosl.org/pipermail/replicant/attachments/20170413/dfda6f51/attachment-0001.asc>


More information about the Replicant mailing list