Open mon opened 2 years ago
I have the same issue. I need to preprocessor nasm files before assembling them. What is the status on this?
The issue seem to be resolved in master branch.
system parameters
while the current meson version (installed using apt) didn't fix the issue.
0.61.2 is fairly old, can put try something from 0.62 or 0.63 (current)?
git bisect says that this reproducer stopped crashing with commit d0a0e04c987ea92473073f9c3017fe648fc59f04. (Released in 0.63.0)
Congrats @dcbaker, you fixed it. :)
EDIT: No, it's not fixed, building results in the error mentioned above:
$ ninja
ninja: error: 'blank.g1', needed by 'b.out.p/blank.g2', missing and no known rule to make it
I’ll look at it, probably isn’t too hard to fix
If issue is not a high priority me and @duckflyer will be happy to try and work on it.
Oof, this issue is actually a lot harder than it looks. We basically assume that a Generator output is part of a target, and generate the output directory for the generator in the backend based on the conusming Target.
I'm also affected by the issue. I need to use two code generators. The first one is re2c
that will generate lexers. Once I have the output for that, I pass it to a second generator (gperf). However the issue I get from the command line is very different from the initial report for this issue:
ninja: error: 'actor.cpp', needed by 'emilua.p/bc2_actor.cpp', missing and no known rule to make it
File src/actor.ypp
was fed to re2c
to generate actor.cpp
. actor.cpp
should be used as input in gperf
.
Describe the bug Trying to chain generators raises an AssertionError. Patching this code then produces an invalid ninja.build
To Reproduce For this toy example, all that is required is this
meson.build
and a blank file,blank.txt
.output
Modifying the code here so that
GeneratedList
is treated identically toCustomTarget
starts producing invalid ninja depedencies:I've got halfway to a solution using a new sentinel type which is put into the target's private directory during ninja.build generation, but it feels like a very heavy solution. I am dedicated to fixing it but would like some guidance on easier strategies.
system parameters
master
at time of submission - 95a4c6a62a925fc888e7c07ae3d9937c87e1c759