Distro packages

[pombredanne]

We should provide and eventually maintain distribution packages as RPMs and Debian packages.

Beside the basic packaging, this would mean also contributing the package to the distro and addressing some issues to make this particular such as:
#469

[v27u27]

Hello Sir, as per my understanding, are you asking to make debain and RPMs distribution package for the scan code toolkit, so that linux users can directly install this toolkit onto their system ?, Please correct me if i am wrong anywhere.

[pombredanne]

@v27u27 Yes this is exactly that. Note that eventually there could be some tools to help build skeletons for these.

[sudeepb02]

Hello @pombredanne,
I'd like to work on this issue as my GSoC project'17 as it is listed here GSoC 2017 nexb/aboutcode wiki

[pombredanne]

Great! Check also #469 and http://Appimage.org that would be a great candidate for Linux packaging.

[sudeepb02]

@pombredanne Thanks. I'll go through the Appimage documentation and start working on it ASAP.

[shivakrishna9]

Hi @pombredanne I would like to give my first PR for this issue as part of gsoc ideas list.

[pombredanne]

@shivakrishna9 sure, that's great.

[shivakrishna9]

@pombredanne please clarify if this project is assigned to me so that I will be able to completely dedicate my time to this. Thanks.

[pombredanne]

@shivakrishna9 I cannot assign to you anything for now:

1. the GSoC has not even started in earnest. You have not suggested a plan for your project submission. And then you may or not be selected. This will happen in April... I cannot tell anything now.

2. I do not know you. I will get to know you through PRs. This is part of our selection process.

So if you want to submit some PRs (which is a requirement if you plan to submit a project application with our org) this is great. I will welcome this and review it then and when you submit this.

[sudeepb02]

@pombredanne I went through the AppImage documentation for distro packages and made an AppImage for a applications. But I see that most of the work here is testing rather than coding.
I was able to run the AppImage successfully on my system but running it on different distros raise some issues which needs the source to be changed.
So can I take this up as a project for GSoC? As the duration for GSoC is 3 months, I'll also create AppImages for AttributeCode and TraceCode

[pombredanne]

@sudeepb02 Great! Can you point me to a public place where you pushed this?
Now this is a GSOC project alright, but to be complete it should cover in order of priorities:

1. Fix De-vendor prebuilt binaries to ease packaging for Linux distros #469 and this will require quite a few code changes such that its works in scancode proper and the scancode-toolkit-contrib too.

2. Create a build loop that handles all the automated compilation of the natives deps on linux/mac and windows . There is some elements of this in https://github.com/pombreda/thirdparty and https://github.com/pombreda/thirdparty-manylinux/ for the python side. And the native code is listed here: https://github.com/nexB/scancode-thirdparty-src and there are some build notes to help.

3. Once we have this in place, create a proper AppImage AND a test suite to validate this is working on may linuxes (may be using OpenSuse OBS, openshift, Travis, circle, etc.) possibly using container images for this

4. create also proper Debian and RPM packages. Create tests for many distros to smoke check the installation. Submit the packages for inclusion in the distros.

5. also create packages and/or ports for other places: e.g. homebrew, archlinux, Guix, FreeBSD

6. also create packages for other languages wrapping ScanCode to run it in a subprocess and get the results back for JavaScript, Java, Ruby. Create proper NPM, Maven and Gem packages and push these to their respective repos . Consider also creating a Maven plugin to scan while you build. Consider other build plugins for similar tools, e.g. Jenkins.

7. do similar things for other tools such as AboutCodeMgr and TraceCode.

8. to 4. are must have. 5. to 7. are bonuses if time permits .

The creation and tests for all this must be automated end-to-end... so I guess this can fill up three months alright. But if this is not the case, there will be for sure a lot of things in that domain that can fill any voids and slack...