Problems building Subsurface for Android

Dirk Hohndel dirk at
Thu Feb 2 12:47:48 PST 2017

On Thu, Feb 02, 2017 at 09:35:01PM +0100, Anton Lundin wrote:
> On 02 February, 2017 - Dirk Hohndel wrote:
> > On Thu, Feb 02, 2017 at 09:10:36PM +0100, Anton Lundin wrote:
> > > > 
> > > > > Goes to show you that whatever we do, we always make assumptions.
> > > 
> > > Yea, everything is based on lots of assumptions. I wrote that script to
> > > make it possible for others than me to build the apk's , and at least 
> > > both Dirk and Jocke have managed to build apk's so it can't be that
> > > broken.
> > 
> > It isn't. It just doesn't document all of its assumptions
> I thought most if it was documented in packaging/android/README . It
> could check its prerequsits and give better error messages.

I think that was all I was talking about. And no, I wasn't pointing
fingers at you. I have written my share of these scripts and I constantly
find yet another assumption that I have made.

> > > And for everyone who tries to write a shellscript, you might think you
> > > know what you're doing, and you most probably don't. Please use
> > > before you end up doing rm -rf / directory.
> > 
> > I suck at writing shell scripts. More than I suck at writing QML. That
> > said, what else should I be doing in order to make things easier for more
> > people to contribute?
> Use shellcheck, centralize and don't over-use shell. Those are my
> recommendations about shell.

I'll need to use shellcheck for sure.
The challenge with "centralize" is that the different scripts are already
quite fragile, combining them into one may seem like a good idea, but I
think it would likely create one script that is even more fragile.
But in the end I'm sure you are right. I just don't think that I'm the
best person to do that :(

> We can't solve everything with creating another tool. That way we soon
> have our own subsurface-developer-distro, which you might need a whole
> on-prem-cloud to run.


> If our documentation is wrong, out of date, or just hard to read, we
> should fix that. If the tools we use gives non-explanatory error
> messages, we should fix those. If some people need help and guidance, we
> should try to provide that.


> If you can write and I would be
> impressed, but I hope no one expects you to.

They do. They definitely do.


More information about the subsurface mailing list