Build our own 0.9.8 on travis

[reaperhulk]

This is an experiment to stop using the lucid hack and custom compile OpenSSL 0.9.8zh for coverage purposes on travis.

[mention-bot]

By analyzing the blame information on this pull request, we identified @alex, @Ayrx and @dstufft to be potential reviewers

[alex]

This also removes several builds, can we make removing builds and switiching how we compile OpenSSL be two separate steps?

[codecov-io]

Current coverage is 99.98%

Merging #2553 into master will not affect coverage as of 2f044a2

@@            master   #2553   diff @@
======================================
  Files          125     125       
  Stmts        13700   13773    +73
  Branches      1465    1472     +7
  Methods          0       0       
======================================
+ Hit          13698   13771    +73
  Partial          2       2       
  Missed           0       0       

Review entire Coverage Diff as of 2f044a2

Powered by Codecov. Updated on successful CI builds.

[reaperhulk]

Coverage problems are due to being newer than 0.9.8m, when we have some branches based on that. I'll drop us to 0.9.8l I guess (yuck).

[reaperhulk]

and 0.9.8l doesn't want to compile. going to let this one sit for a bit while I think about what to do...

[alex]

Can we add the letter here, in the event we want to add more in the future I don't want untangling this to be a pain.

[alex]

Besides that comment this looks ok to me, but I'd like another pair of eyes because this is a little wild.

Also, how long does compilation take? Keep in mind it will happen on the first push in any PR series.

[reaperhulk]

It adds about 2-3 minutes to the runtime of the job within the build.

[alex]

👍

[alex]

Unless there's a compelling reason can we drop the v, it makes the logs hard to read.

[dstufft]

This change looks OK To me, though I agree with the dropping of v from the tar command.

[coveralls]

Changes Unknown when pulling 80b1281 on reaperhulk:098-change into ** on pyca:master**.