@bagder With the benefit of hindsight, what functionality would you rather not have implemented in curl, or not in the way it currently is, but it’s too late to change now because it’d cause breaking changes?
(I’m assuming something like that exists in curl because the probability of it happening in any software project increases with longevity and popularity, and curl has both)
@vitor there are several smaller and larger choices we now know were less clever but in the name of not breaking things for users we keep them. But I don't think any of them are major nor have any past silly decisions ever put and serious hurdles in the way for taking curl further. By shear luck and some skills.
@adnan if there is a quality open source implementation we can use, I feel no obligation to reinvent the wheel, no. This is how we do HTTP/3 (four different backend options)
@bagder
"If you made a list of all the basic maintainer task, the one that need to be done to just keep things going. What would they be and which one has the worst tooling?"
Note that i do not mean answering patch email and all, I mean things like keeping the dependencies up to date and the build system running.
@bagder (definitely not fishing for a particular answer here and definitely not having something cooking for which i want data to prove a point about hobbyists maintainers and "supply chain")
@Di4na I think it is hard to answer that accurately as I don't really measure a lot of these things. I work on curl stuff I think needs to be done, maybe 50-60 hours per week - most of that without keeping tabs or tracking what I spend the time on. I suspect the boring parts might feel larger than they are.
@bagder I mean this is a fair answer. And it does provide data. It shows we could benefit from looking at this, to at least understand better what would have impact... Or if this idea have no impact at all.
Thanks! It is the answer I expected.
Only follow up question. Is it more than 2h per month?
@Di4na just the specific keeping the build system up and running is more than 2h per month for sure. For curl, keeping track of and having dependencies up to date is not a major task as we need to work with a wide variety of versions. But we spend significant time on making sure curl works with all those different versions of our bazillion possible dependencies. Way more than 2h per month..
@bagder right. Which already would be a strain on a "hobbyist sometimes on weekend" schedule.
I asked because I am trying to find ways to communicate about the realities of maintainership on these schedules. And maybe see if I cannot offer a view that may lead to a more impactful action plan than "give money to everyone". Not that money does not help sometimes or would not be part of it buuuut...
Add comment