Ok, so I ran into a problem because rusage.h has been renamed pike_rusage.h due to old dependencies. Thus I run "make depend". However this fails as well, because make depend compiles Pike. This is no good - make depend should make dependencies! I have to manually edit the 'dependencies' files.
How about using the method used by PExts instead - the configure process creates the dependencies. Works really well I think (does of course make configuration slower).
Uh, isn't the question of whether compiling pike is necessary to make depend rather orthogonal to wherher the dependecies are created from the configure script or the Makefile? I prefer to keep it in the Makefile since it makes it easier to remake the dependenices, but I agree that `make depend' shouldn't try to compile pike first.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-02-19 20:38: Subject: make depend (7.5 cvs)
Ok, so I ran into a problem because rusage.h has been renamed pike_rusage.h due to old dependencies. Thus I run "make depend". However this fails as well, because make depend compiles Pike. This is no good - make depend should make dependencies! I have to manually edit the 'dependencies' files.
How about using the method used by PExts instead - the configure process creates the dependencies. Works really well I think (does of course make configuration slower).
/ David Hedbor
Especially because:
Skipping make depend in post_modules since a fresh pike is needed to make the autogenerated sources there.
(seems like it built tpike btw)
/ David Hedbor
Previous text:
2003-02-19 20:42: Subject: make depend (7.5 cvs)
Uh, isn't the question of whether compiling pike is necessary to make depend rather orthogonal to wherher the dependecies are created from the configure script or the Makefile? I prefer to keep it in the Makefile since it makes it easier to remake the dependenices, but I agree that `make depend' shouldn't try to compile pike first.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
?PARSE ERROR READY.
Are you saying that dependenices should not be generated for post_modules? How about having a separate target `post_depend' that makes depend for the post_modules, after the pike has been compiled?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-02-19 20:43: Subject: make depend (7.5 cvs)
Especially because:
Skipping make depend in post_modules since a fresh pike is needed to make the autogenerated sources there.
(seems like it built tpike btw)
/ David Hedbor
'make depend' first compile tpike (including all modules), and (I assume) made dependencies for all sub-modules (this did of course require manual removal+touching or editing of the old dependencies first). When it comes to "post_modules" it exits because there isn't a new "pike". So I type make. It then failed at post_modules because the "dependencies" were missing (this might have been a fault on my end though, not sure).
In either case, annoying and definitely nothing I've really noticed before.
/ David Hedbor
Previous text:
2003-02-19 20:47: Subject: make depend (7.5 cvs)
?PARSE ERROR READY.
Are you saying that dependenices should not be generated for post_modules? How about having a separate target `post_depend' that makes depend for the post_modules, after the pike has been compiled?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
There's a note about that in README-CVS. Removing tpike should go a long way in avoiding this awkwardness. Probably all of it.
/ Martin Stjernholm, Roxen IS
Previous text:
2003-02-19 20:51: Subject: make depend (7.5 cvs)
'make depend' first compile tpike (including all modules), and (I assume) made dependencies for all sub-modules (this did of course require manual removal+touching or editing of the old dependencies first). When it comes to "post_modules" it exits because there isn't a new "pike". So I type make. It then failed at post_modules because the "dependencies" were missing (this might have been a fault on my end though, not sure).
In either case, annoying and definitely nothing I've really noticed before.
/ David Hedbor
It does; tpike is no more. The depend target already had a provision for not entering post_modules when a fresh pike doesn't exist.
/ Martin Stjernholm, Roxen IS
Previous text:
2003-02-19 23:12: Subject: make depend (7.5 cvs)
There's a note about that in README-CVS. Removing tpike should go a long way in avoiding this awkwardness. Probably all of it.
/ Martin Stjernholm, Roxen IS
That is the same problem I ran into a while ago. The current solution is appearantly to do "make srcclean" or something to remove all the generated files, including the dependency files.
I think that make depend always should succeed and not be dependant on Pike. If there is parts of the depend tree that depends on Pike, and Pike isn't built yet, and the system Pike is worthless, those dependency files should at least be cleared, or marked so that they get remade when Pike is built.
/ Mirar
Previous text:
2003-02-19 20:38: Subject: make depend (7.5 cvs)
Ok, so I ran into a problem because rusage.h has been renamed pike_rusage.h due to old dependencies. Thus I run "make depend". However this fails as well, because make depend compiles Pike. This is no good - make depend should make dependencies! I have to manually edit the 'dependencies' files.
How about using the method used by PExts instead - the configure process creates the dependencies. Works really well I think (does of course make configuration slower).
/ David Hedbor
I think that with the exception of post_modules, it should be possible to generate proper dependencies without having a working pike. For cmods, it's just a question of running gcc -MM -MG on the .cmod instead of on the generated file, since they contain the same #include directives.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-02-19 21:10: Subject: make depend (7.5 cvs)
That is the same problem I ran into a while ago. The current solution is appearantly to do "make srcclean" or something to remove all the generated files, including the dependency files.
I think that make depend always should succeed and not be dependant on Pike. If there is parts of the depend tree that depends on Pike, and Pike isn't built yet, and the system Pike is worthless, those dependency files should at least be cleared, or marked so that they get remade when Pike is built.
/ Mirar
pike-devel@lists.lysator.liu.se
-
David Hedbor ＠ Pike developers forum -
Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum
-
Martin Stjernholm, Roxen IS ＠ Pike developers forum
-
Mirar ＠ Pike developers forum