Makefiles are driving me batty

[dcon]

I have a makefile that has

resources: myexe
	python3 <script that creates a MO files>
	python3 <script that creates a zip file with said MO files>

all: resources myexe
	python3 <script to run myexe>

If I do a make resources, all is good. But if I just make (or make all), it looks like I get some ancient version of the zip. I put all sorts of debugging into the python script and I see the files get added. But they just don't exist in the zip.

I've even searched the system with a find / -name ... and there just ain't nothin there!

(And, yes, even after a make resources, a make will put back the old file.)

I'm losing what little hair I have left.

[Gribnit]

@dcon have you tried batch files? Or make clean first?

[dcon]

@Gribnit said in Makefiles are driving me batty:

@dcon have you tried batch files? Or make clean first?

I'm on Ubuntu (and MacOS). And I've completely deleted the build directory.

[Tsaukpaetra]

@dcon said in Makefiles are driving me batty:

<script that creates a zip fil </blockquote>

Verbosify that, see where it thinks it is, where it thinks the files are, etc.

[dcon]

Oh, even better... I stuck a cp zipfile newname in there. newname is correct. zipfile is wrong.

[dcon]

@Tsaukpaetra said in Makefiles are driving me batty:

Verbosify that, see where it thinks it is, where it thinks the files are, etc.

I did make -d >out - looking for 'resources', things look proper. And I only see the actual filename where the script spits it out.

[Tsaukpaetra]

@dcon said in Makefiles are driving me batty:

I only see the actual filename where the script spits it out.

Is the script immutable?

[dcon]

It was a different script. One that was only supposed to be running tests, not modifying things.

(found it by adding a chmod 400 in the makefile - that other script then failed)

And I'm all out of liquor...

[HardwareGeek]

@dcon said in Makefiles are driving me batty:

It was a different script.

BTDT

[dkf]

@dcon said in Makefiles are driving me batty:

The surest sign that you're working with build automation is a series of new dents in your desk.

[PleegWat]

@dkf The only thing worse than make is all of the alternatives?

[dkf]

@PleegWat make works well… provided you can follow the “this file generates exactly that file” paradigm that is common in C and C++. It really sucks for Java (BTDT) because Java has a simultaneous-declaration model, and hence likes to recompile the world whole project at once. (Fortunately, that's quick.) make is also not a very good fit for (La)TeX, but there it's because you potentially have to repeatedly run the build in a loop until it stabilises the cross-references, and there's potentially no upper bound on how long that takes.

In both cases, you can make it work… sort of… but it's horrid and you're better off doing something else.

[PleegWat]

@dkf We use a utility to generate function declarations automatically. It has been suggested to run this utility automatically during the build phase. I've always been opposed to that because I have no idea how that could work without always rebuilding all object files.