Add support for Emacs-25.0.94 pre-test

[phillord]

As discussed in #69.

I've had to refactor some parts -- particularly I've removed the "z" from "tar xvfz". My tar just copes with this (and does xv files also). Don't know about the mac tar though.

[rejeep]

1. This only adds support for a compiled version and not a Travis version.
2. I'm not sure about calling it emacs-25.0.94. What happens when there's another prerelease? Do we create another recipe then? If we call it emacs-25-prerelease, we only need to update the path to the tar-ball and build a new travis version.

I also wonder what the real purpose of this is, specially as you didn't add a Travis version?

[phillord]

Addressed most of these -- not sure how you work out the URL for the travis version.

Agree about the "prerelease" tag. Decided one way, then changed my mind, but decide that this is probably best.

[rejeep]

Well, it's a little bit of work every time a new Travis version is added, that's why I'm a big sceptical about it. At least I have the procedure documented: https://github.com/rejeep/evm#adding-travis-binary

[phillord]

I don't mind making the binaries, in so far as I have permission.

I'm not sure that I understand the instructions for generating the binary. What is the purpose of vagrant here? Can the binaries not just be produced in a host linux box? Also, why is the make bootstrap necessary -- the make install should do this anyway but make bootstrap will force recompilation of the elc files (which is pretty slow).

[rejeep]

What is the purpose of vagrant here? Can the binaries not just be produced in a host linux box?

As long as they are compiled on the same architecture as on Travis I guess it's ok. But since Travis provides a very simple way of setting up an exact copy of their environment, why not use it?

Also, why is the make bootstrap necessary

I usually compile from a new tar-ball every time I compile Emacs and then it doesn't matter I guess.

I may be wrong, but doesn't the INSTALL say bootstrap should be used?

[phillord]

make bootstrap is a quite a bit slower because the src tar ball normally contains the elc files. It's good when bisecting but otherwise, not really needed.

I build a binary tonight and put it somewhere useful!

[rejeep]

make bootstrap is a quite a bit slower because the src tar ball normally contains the elc files. It's good when bisecting but otherwise, not really needed.

Then I guess recipes where Emacs is compiled from scratch could be updated, for example: https://github.com/rejeep/evm/blob/master/recipes/emacs-24.5.rb#L16-L18

I build a binary tonight and put it somewhere useful!

I store all pre compiled binaries under each release, see: https://github.com/rejeep/evm/releases

[phillord]

Think that's it!

[phillord]

Just pinging. Think this is complete now. I have added a binary on http://www.russet.org.uk/scratch.

[rejeep]

@phillord Sorry for the delay, I have very little time left over with a six month old at home... 😕

I made a few comments, please adress those and I will try to get this in production as soon as possible.

[phillord]

@rejeep No worries about the delay. Having a six month at home is cause for celebration.

I've made all the changes. I have checked -- make bootstrap does not seem necessary (and the compile is far quicker, since no byte compilation is needed).

[rejeep]

@phillord All done now, wanna try it out please!?

[phillord]

Alas, no, it is not working.

Something strange is going on. I am getting three copies of Emacs

https://travis-ci.org/phillord/assess/builds/136210347

which does find /tmp -type d

There are two copies of Emacs-24.3 installed by cask, one at /tmp/emacs-24.3 and one at /tmp/tmp/emacs-24.3. And Emacs 25 has been installed at /tmp/tmp/emacs-25-pre. Oh dear.

[rejeep]

I guess you are using https://gist.github.com/rejeep/ebcd57c3af83b049833b#file-travis-evm-cask-sh-L22?

one at /tmp/tmp/emacs-24.3

This is where the tar-ball in unzipped and compiled.

one at /tmp/emacs-24.3

That's the real installation path from the script above (--prefix)

And Emacs 25 has been installed at /tmp/tmp/emacs-25-pre

There should also be a /tmp/emacs-25-pre

[phillord]

Okay, checked. So, it appears that you were entirely correct -- the binaries are not working on travis. So, the whole vagrant step is necessary.

Oh dear, my bad. I will try and get this bit working over the next few days. Apologies for time waste.

[rejeep]

No worries, let me know when you are done.

[phillord]

New binary is here:

http://www.russet.org.uk/scratch/emacs-25-pre-travis.tgz

It's actually 25.0.95 as there has been a new pre-test in the meantime. I've worked out how to test it now, and you can see an Emacs hello world here:

https://travis-ci.org/phillord/evm-tester

Again, apologies for making you do this twice!

[rejeep]

Updated, can you please try again!?

Again, apologies for making you do this twice!

No worries, just glad I don't have to do the compiling myself... 😄

[phillord]

Ack --- I the used wrong extension on the tarball -- .tgz not .tar.gz. So, it 404s at the moment.

[rejeep]

Would it work to simply rename it?

[rejeep]

I renamed it

[rejeep]

@phillord I added you as a collaborator to the repo

[phillord]

Should do, yes. It's just the name.

On Mon, June 13, 2016 8:40 am, Johan Andersson wrote:

Would it work to simply rename it?

---

You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: https://github.com/rejeep/evm/pull/70#issuecomment-225509057

[phillord]

All sorted -- it's working well now. And accepted collaborator.

Once Emacs 25 is released, I'll think about doing a build from master branch if you are happy.

[syohex]

Emacs 25.1 is released now.

[phillord]

Yes. I'll make these binaries over the next couple of days.

[rejeep]

See https://github.com/rejeep/evm/pull/73