Prefix output for sub-make phony targets? - makefile

Question

Problem
Given the following Makefile:
.PHONY: blah blah1 blah2
blah:
#$(MAKE) -j --no-print-directory blah1 blah2
blah1:
#echo "hello"
#caddy run
blah2:
#echo "goodbye"
#esbuild --watch
Running make blah returns the following:
❯ make blah
hello
Caddy listening on :3000
goodbye
esbuild building...
esbuild complete in 4ms
How can I prefix the name of the target before any output (stdout or stderr) from that target?
I'm looking to get this:
❯ make blah
[blah1] hello
[blah1] Caddy listening on :3000
[blah2] goodbye
[blah2] esbuild building...
[blah2] esbuild complete in 4ms
Version
GNU Make v4.4
Other solutions I've tried
I can hack this together by manually appending the output through sed on each line inside the target definition:
.PHONY: blah blah1 blah2
blah:
#$(MAKE) -j --no-print-directory blah1 blah2
blah1:
#echo "hello" | sed 's|^\(.*\)$$|[blah1] \1|'
#caddy run 2>&1 | sed 's|^\(.*\)$$|[blah1] \1|'
blah2:
#echo "goodbye" | sed 's|^\(.*\)$$|[blah2] \1|'
#esbuild --watch 2>&1 | sed 's|^\(.*\)$$|[blah2] \1|'
... but this solution gets very tedious very fast for non-contrived Makefiles.

Answer 1

If you want to manipulate the output of your recipe's you're going to have to pipe them to some sort of script (make doesn't offer anything like that). That being said, you can shorten it a bit:
.PHONY: blah blah1 blah2
blah:
#$(MAKE) -j --no-print-directory blah1 blah2
blah1:
#echo "[$#] hello"
#caddy run 2>&1 | awk '$$0="[$#]"$$0'
blah2:
#echo "[$#] goodbye"
#esbuild --watch 2>&1 | awk '$$0="[$#]"$$0'
Now, your question also suggests you were looking for something that is sub-make specific (as opposed to target specific) so, if that's the case, you can simply pipe the output of sub-make to your script:
.PHONY: blah blah1 blah2
blah:
#$(MAKE) -j --no-print-directory blah1 | awk '$$0="[blah1]"$$0'
#$(MAKE) -j --no-print-directory blah2 | awk '$$0="[blah2]"$$0'
##note: -j blah1 and blah2 are build sequentially here
blah1:
#echo "hello"
#caddy run 2>&1'
blah2:
#echo "goodbye"
#esbuild --watch 2>&1
--- EDIT ---
You could run two different instances in make in parallel, and pipe the output of those through the prefix script as so:
.PHONY: blah blah1 blah2 submake_blah1 submake_blah2
blah: submake_blah1 submake_blah2
submake_%:
#$(MAKE) -j --no-print-directory $* | awk '$$0="[$*]"$$0'
blah1:
#echo "hello"
#caddy run 2>&1'
blah2:
#echo "goodbye"
#esbuild --watch 2>&1
This of course involves you invoking your parent make with -j to allow these makes to run in parallel. Also, if the two submakes have common dependencies (i.e. blah1 depends on blah32, and blah2 also depends on blah32), then this would result in blah32 being built twice, and could result in race conditions.

Answer 2

As Henry shows it's tricky to do what you asked.
However, it could be that this is an XY problem and what you REALLY want to be able to do is just figure out which output comes from which target.
In that case you should investigate the output sync facility provided by GNU make, which collects the output of the target and generates it all in a single block, so that it's not mixed together with other targets' output. For example something like:
.PHONY: blah blah1 blah2
blah:
#$(MAKE) -j -O --no-print-directory blah1 blah2
blah1:
#echo "start hello"
#caddy run
#echo "end hello"
blah2:
#echo "start goodbye"
#esbuild --watch
#echo "end goodbye"
This won't prefix every line with the target being built but it will group all the output for each target together in the output, rather than mixing it all up.