- From: Norm Tovey-Walsh <ndw@nwalsh.com>
- Date: Tue, 14 May 2024 13:09:30 +0100
- To: denis.maier@unibe.ch
- Cc: xproc-dev@w3.org
- Message-ID: <m2wmnw4mcw.fsf@nwalsh.com>
> But should it be possible to use xproc pipelines instead of a makefile? You can, but there are tradeoffs. Tools like Make (and Gradle, etc.) are designed to do some level of dependency management for you. If A depends on B and B depends on C, then a change to C requires rebuilding A and B. But if B hasn’t changed, you don’t need to rebuild A. Out of the box, XProc doesn’t have that kind of functionality. (I wrote it for my 1.0 processor years ago and will likely re-implement it in my 3.0 processor in the fullness of time, but it’s not standard functionality.) On the other hand, if you’re using Make, you’re potentially starting a new JVM process for every step. That can be expensive. (Years ago, I got heaps of praise from a user who cut a big job down from 11 hours to tens of minutes by switching to XProc. The big win wasn’t really XProc, exactly, it was not starting a new JVM instance hundreds of thousands of times.) If the pipeline is easier to maintain and you don’t find its performance problematic, you win! :-) Be seeing you, norm -- Norm Tovey-Walsh <ndw@nwalsh.com> https://norm.tovey-walsh.com/ > Any sufficiently abstract art form is indistinguisable from > rubbish.--Sean McGrath
Received on Tuesday, 14 May 2024 12:19:36 UTC