Repository navigation
conditionally cache generates keys #2876
Description
Description
I would love if taskfile can not only conditionally perform a task, but by telling it what it generates that it would be able to cache everything that was generated by a chechsum so it can copied over rather than re-execute.
I understand though that first #2181 needs to be addressed.
Activity
trulede commentedon Jun 8, 2026
For that you should consider a cache solution. Most compiler toolchains will already support some form of caching, or you can easily create one yourself - a
deptask that looks in the checksum file and, if available, downloads files from a cache (based on the checksums).As cache strategies are going to be very domain specific I can't imagine a generic solution in Task making too much sense (please prove me wrong ....). Better just write a
deptask to do whatever you need.Take a look at the checksum files and decide if there is enough info available to implement your own cache.
hanoii commentedon Jun 9, 2026
I hope you don't take this in a wrong way, as my intention is being constructive. This is the third issue is a short time you replied to me with something in the like of: You can easily do X on a task Y. I don't think it's an useful comment. I know that. You can easily do a lot of what task do as well. What changes is where maintainers want to draw the line.
Having said that, I offer ideas or feature requests of things that I've already done myself in different ways. But when I find a tool that's has the potential to be ubiquitous in my dev processes and one I can recommend or enforce, and specially those that already have something built closer to it I like to verbalize it.
I now sometimes it could be scratching my own itch, but a lot of times it might be a good idea for me, the maintainer or anyone else to implement.
My use case:
If I can tell task to keep a cache of generated files (I am already configuring the generated files I want it to look after), maybe specified something other things like access times or expire date or anything in the like I can have this already there and ready to be used whenever I want.
Also, this particular cache strategy is not that very tied to the domain. Is a hash -> artifact cache strategy. It can be used with a lot of different domains.
trulede commentedon Jun 9, 2026
Task should support everything you need to realise your use case.
If you find a limitation that prevents such a use case, then open an issue so that your use case can be enabled (or open a PR!)
However, Task is not a cache service, or a database for metadata. Task is a task runner, and you can write tasks to do those things (which will be far more effective anyway).
(I will convert this to a discussion)