Mathieu Lirzin <m...@gnu.org> skribis: > l...@gnu.org (Ludovic Courtès) writes: > >> Mathieu Lirzin <m...@gnu.org> skribis: >> >>> Anyway, inspired by what is done by Automake for info manuals I have tried >>> to >>> build man pages in $(srcdir), which is fixing the issue too. The bonus is >>> that now all the documentation is consistently built in $(srcdir) and that >>> we >>> avoid adding another "case $? ..." trick. >> >> This sounds reasonable (in fact I wonder why we were not doing it >> already given that things are supposed to live in $(srcdir)). > > Why were you expecting such thing?
Because as you wrote, it’s how the .info target works, and understandably so. >> Did you confirm that ‘make distcheck’ passes after that, and that >> ‘help2man’ is not invoked when building from a tarball? > > I can't confirm for ‘make distcheck’ right now, because of: > > substitute: updating list of substitutes from 'http://hydra.gnu.org'... 33.3% > 4 packages are not substitutable: > /gnu/store/5apc4w376ls7fhydg08plc20agry7pn4-emacs-24.5 > /gnu/store/zfxj52lh6h6q7jmb1w2s34h004zsbaw3-emacs-24.5 > /gnu/store/10qcwd0gxypm977ksbvhjrpa4m1xk334-emacs-24.5 > /gnu/store/r8xphs9ka0lwmwhzxv6mfs9j8nm91843-emacs-24.5 > > Makefile:4913 : la recette pour la cible « assert-binaries-available » a > échouée I’m sure you’ll easily find how to annihilate the ‘assert-binaries-available’. Find another excuse. ;-) > However I have manually checked that with 'make dist' + (regular build or > VPATH build) 'help2man' is not invoked. Good. >> Also, do these changes preserve the benefits of >> 36629097523b0abd89d1b931293150cb9c1f242d (“build: Generate man pages >> after compiling Guile objects.”)? > > Yes it is. That would be unacceptable if not. ;) Wonderful! So we’re all set, aren’t we? Thanks, Ludo’.
