5 ms·
There's been a substantial shift recently, in cloud deployment anyway, where the work entails stitching together many FOSS tools. I think this started as a con
by imglorp 7y ago
There's been a substantial shift recently, in cloud deployment anyway, where the work entails stitching together many FOSS tools. I think this started as a concept around 2000 with the idea of LAMP stacks and kept going My last several jobs, and everyone around me, have amounted to duct tape.
In this kind of situation, there's plenty of room--sometimes company sanctioned even--to spend time improving the tools and contributing the changes back. Win-win-win for everyone.
So not explicitly FOSS jobs, but it's everywhere already.
- pmjordan 7y agoit's everywhere already If only everyone using the FOSS tools actually contributed back their local modifications. That's definitely not happening "everywhere".
- chrisseaton 7y agoWhat are people doing? Maintaining extensive long-term private forks of basic infrastructure tools outside of their business line and core competency? Why would anyone choose to do that?
- cydonian_monk 7y agoCultural reasons. Legal reasons. And sometimes (rarely) the changes you might make to these basic infrastructure tools are of a proprietary nature. It's frustrating at times, but mostly it comes down to whether management or the legal department thinks you should be allowed to contribute back.
- jlg23 7y agoIn my experience: Monkey patching when in maintenance mode. "Yes, for historical reasons we have to pipe every incoming mail to this collection of awk, perl, ruby, python and brainfuck scripts."[1] [1] Unless you work for German Telekom, one language is a lie.
- gravypod 7y agoI've tried to upstream changes or bug reports to open software. Sometimes something that is a bug to you is a feature to someone else.
- pmjordan 7y agoIt's also not uncommon that a 10-line local hack that took an hour or two total to figure out and implement mutates into a 200-line pull request that eventually gets merged after 3 months of near-daily babysitting and 7 rounds of review with 10-deep email threads because the initial approach wasn't quite up to standard and the project maintainers have meant to redo the code in this area for a while now anyway, so could you just… Or, you know, somewhere along that path, you give up, for reasons which can be anywhere between frivolous/laziness and entirely reasonable. If you're lucky, your change is sufficiently googlable that one of the next people who need it comes along and cleans it up and goes through with the full merge dance before upstream has diverged too far from your abandoned patch.
- pmjordan 7y agoYou give people far too much credit. In my experience: implement a hack in whatever the latest upstream version is to meet requirements, then never pull in upstream changes and just keep running the increasingly ancient, hacked-up fork.
- Ididntdothis 7y agoIs the contributing back part actually happening? To me it seems a lot of the cloud stuff takes open source code and puts a proprietary wrapper around it.
- imglorp 7y agoAh, different use case than what I was thinking (integration). I think you're talking about offering the FOSS as its own managed component like AWS hosting Redis. One would hope they post back patches, but they are getting pushback like the Mongodb dustup last year: https://www.geekwire.com/2019/mongodbs-licensing-changes-led-red-hat-drop-database-latest-version-server-os/ https://www.geekwire.com/2019/mongodbs-licensing-changes-led... etc