Re: [boost] [EXTERNAL] [mpl] [build] [testing] Merge pull requests

On Sep 19, 2014, at 12:27 AM, Andrey Semashev wrote:

On Friday 19 September 2014 03:54:47 Belcourt, Kenneth wrote:

...

On Sep 17, 2014, at 12:39 PM, Andrey Semashev wrote:

...

There were a lot of discussions about MPL recently, and I created a few pull requests to address issues:

https://github.com/boostorg/boost/pull/39

Invoke 'headers' target before building testing tools. Note that several testers are currently failing to run because of this problem. Since this affects the testing process, I'd appreciate if someone familiar with it could take a look.

I’d like to revert the MPL PR that caused our testing problems, and the one Edward just applied. It was wrong of me to commit the PR without any testing. There’s been some list discussion but my overall concern is that we can’t clearly explain why some develop testers are broken and others are working okay. The commit completely broke Darwin, Debian, FreeBSD, QNX and Windows. But what’s really worrying is that some RHEL testers are still merrily cycling while other RHEL testers are broken. Same for Ubuntu, some are cycling okay while some are broken. I think it’s too risky to apply a patch when we, honestly I, don’t understand why this is happening.

While we could apply Andrey’s PR (below) I’m afraid that may mask real underlying problems that we may not find out about until this hits master, or worse, users. I’ll work with Andrey to get the MPL changes applied, perhaps in smaller PRs.

Please let me know if you have concerns.

Kenneth,

dba Noel

I've already explained why the problem has likely appeared, and moving back to the monolithic MPL is what covers it. It's not the fix.

We’re in agreement that implicit-dependency is broken for some reason. But note that several of our internal projects utilize BB implicit-dependency and it works fine for us. My concern is that these projects will break when they pull Boost with this change, and this may well happen to other users.

I admit I don't know why implicit-dependency doesn't work in the tools Jamfile, but it is clear that the intention was to invoke 'headers' prior to building.

Yes, but be aware that it’s not just tools, every invocation of b2 is broken, fails to install full set of headers.

That's what my second PR achieves, and I think this is the correct fix. Did you try it?

Yes, it works on Linux, but I’m still reluctant to apply it without a firm understanding why the random tester behavior is occurring. I’m sort of okay with wholesale failure but I’m not okay with some RHEL and Ubuntu testers working okay, others failing.

The problem with implicit-dependency should be investigated too, and Vladimir indicated in the PR that he wants to do that.

Yes, saw that.

You can revert MPL if you still want to, but I honestly don't know how else (on the MPL side) we can do this.

I guess that’s really the question I plan to investigate (while the develop testers are all working correctly).