Hey all, I noticed updating `instructions` on a Do...
# generalr
ripe-gigabyte-88964
10/03/2023, 2:54 PMHey all, I noticed updating
instructions--changed-sincel
late-advantage-75311
10/03/2023, 3:24 PMlate-advantage-75311
10/03/2023, 3:24 PMEvery target in a BUILD file is changed when it's edited.
r
ripe-gigabyte-88964
10/03/2023, 3:25 PMHmm, I wonder if this
docker_imagel
late-advantage-75311
10/03/2023, 3:26 PMoh, so there is no git diff on the build file itself?
r
ripe-gigabyte-88964
10/03/2023, 4:05 PMcorrect
c
curved-television-6568
10/03/2023, 4:33 PMI think this is ~related to this issue https://github.com/pantsbuild/pants/issues/19805
curved-television-6568
10/03/2023, 4:35 PMand as you already suspect, when changes are made to a file that is not linked to the target in any way pants know of (in terms of dependencies and the targets BUILD file), changed-since won’t work.
curved-television-6568
10/03/2023, 4:36 PMin order to address 19805, my plan is to infer additional dependencies of some sort to all contributing sources that impacts the targets field values. no easy solution, so progress has been slow.
curved-television-6568
10/03/2023, 4:38 PMI imagine a crude work-around could be to have an explicit dependency to the macro file itself from all targets the macro creates..
l
late-advantage-75311
10/03/2023, 6:51 PMThis sounds very challenging for the case of a change in a centralized macros file. I am curious if there is any tracing right now from the target to the macro that aided to create it. I've kind of assumed that the macro just executes as a plain function with no tracing and is in every way equivalent to what would happen if the macro had been inlined into the places it is invoked.
late-advantage-75311
10/03/2023, 6:52 PMAnd Andreas, for the case of 19805, I'll comment in there