Hacker News new | comments | show | ask | jobs | submitlogin
Redo, buildroot, and serializing parallel logs (apenwarr.ca)
27 points by dankohn1 4 months ago | hide | past | web | 10 comments | favorite

If you're looking to serialize parallel build logs without changing to an entirely new build tool:

1. Electric Make, a high-performance reimplementation of GNU make and ninja, has properly serialized build output logs since it was introduced in 2002 (https://electric-cloud.com/products/electricaccelerator) (disclaimer: I'm the chief architect for ElectricAccelerator, of which Electric Make is a component).

2. I published a technique on CM Crossroads you could use with GNU make 3.81 to descramble parallel build logs in 2009. The article has moved around since then but these days it seems to be found at https://www.cmcrossroads.com/article/descrambling-parallel-b....

3. The maintainers of GNU make took the concept described in that article and baked it into GNU make itself in 2013 for version 4.0 (http://git.savannah.gnu.org/cgit/make.git/tree/NEWS?h=4.0)

That is a good article, but it doesn’t seem to address several of the concerns raised in my redo article. In particular, it looks like the final output is still in a nondeterministic order, it doesn’t print output from a given target incrementally while it runs, and you can’t query the log retroactively to look for clues in a very big run.

It’s a big advantage to not have to change tools, of course. Although redo happily interoperates with makefiles (which is how the buildroot patch works) so there’s no need to convert everything to get most of the advantages.

Correct, both the technique described in that article and the feature that eventually wound up in GNU make only disentangle output from concurrently executing build processes. With significantly more work GNU make could probably be made to enforce a deterministic order.

However, Electric Make does emit the output in a deterministic order, in fact in exactly the same order as the build would have produced had it run serially. In practice the delayed output is not such a big deal -- people just don't really seem to care that much, when the build overall finishes 20x - 30x faster than it used to. Electric Make also generates an annotated build log, essentially an XML-marked-up version of the log, which contains a tremendous amount of additional information about the build and vastly simplifies debugging, actually.

Hey, Electric Make looks really cool - I ran into it while I was tinkering with https://github.com/dan-v/rattlesnakeos-stack. I also have done a fair few builds in buildroot. It wasn't clear to me at all if you offer anything for open source work / hobbyists - I'd love to try it out.

> I'm sure almost everyone reading this thinks I'm hopelessly pedantic to care so much about the sequence of lines of output in my build logs.

Not at all! In fact there are other tools in other domains that could benefit from careful thought about how to interleave parallel outputs in a way that favors interactivity/immediate output but preserves (effective) ordering [1][2]. This is a tough problem a lot of tools just give up on, but the status quo of "buffer everything until it is completely done" grates against one's senses.

[1] https://github.com/saltstack/salt/issues/22733

[2] https://github.com/ansible/ansible/issues/3887

Wow, those are quite fascinating use cases I hadn't considered. I'm actually curious about things like ansible; what exactly does it do that you couldn't just do with a bunch of redo scripts? Or alternatively, what if you ignored ansible's parallelism and just ran ansible once per host, from redo?

A lot of devops tools, including Ansible and Salt, have higher-level concepts of ordering across hosts. In some cases this is implicit in the way the tool operates and people just take advantage of it, but in others (like Salt's orchestrate runner) the ordering of operations across hosts is explicitly defined by users. That can be useful for scripting blue/green deployment cutovers, pushing out new load balancer rules, etc.

> I'm actually curious about things like ansible; what exactly does it do that you couldn't just do with a bunch of redo scripts?

That might be more of a philosophical question :-) Most of those tools could be replaced with a sufficiently intelligent set of scripts. I think they get their popularity from providing more convenient syntax, building in parallelism for applying changes across hosts, offering cross-platform ways of doing common admin actions like creating accounts, etc.

But if you look at them the right way, they definitely share a lot in common with build systems, particularly when it comes to managing dependent actions.

Several things can be operated like build systems, and this has come up on Hacker News several times. I, for example, maintain my Debian package repository and my GOPHER site using redo, and use redo to import external software configuration into my service management.

* https://news.ycombinator.com/item?id=14837740

* https://news.ycombinator.com/item?id=14928216

* https://news.ycombinator.com/item?id=14836340

* gopher://jdebp.info/h/Softwares/nosh/guide/external-formats.html

* http://jdebp.eu./Softwares/nosh/guide/external-formats.html

My implementation re-uses the MAKELEVEL environment variable to track recursion, making it simpler in the cases where one has GNU make calling redo, or indeed redo calling GNU make. (BSD make unfortunately uses a different and undocumented variable.)

Redo is one of the very few things that ever made me think it might replace make. This implementation is pretty awesome.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact