Hi Dave, > I think that what's really going on here (in the DVI/PS/PDF example) > is that you have dvi generated ps, and you have pdf generated ps. > > I think that it's a mistake for these two different types of generated > files to have exactly the same directory/name.
I'm thinking of a different way to use make: as a "generic" tool to convert anything to anything, not in a specific projet, but in everyday use. I build thus a "generic" makefile, including instructions about how to convert anything to anything. For example, currently on my linux box, I may type "mm foo.html" in a directory where a "foo.sxw" document exists. Then the alias "mm" will call "make --makefile=..." (my generic makefile), which knows how to convert an OpenOffice document in HTML (in this case, by calling an OOo macro). That use of make is a different use than the maybe more classical use "within a specific project", where one knows when designing the makefile what possible goals will be asked. In my generic makefile, I'd like my makefile to accept as much possible goals as possible, and, when asked to produce something, search among the defined rules if it knows how to produce it using anything existing in the folder where I am currently, then do it. > > If I were to create the DVI file, then create the PS, then create the > PDF, edit the DVI and then generate the PS from the PDF, you're now in > a situation where your PS file is out of date but there's no way for > make to know it. Indeed, we can think of strange things happening with such generic makefiles. But we can also think of strange things happening with normal makefiles (I mean, makefiles used more classically, in the specific context of a project). When the user do strange things, the program he is using also behaves strangely. That's normal. If I want a PDF and I'm working with a DVI file (suppose I'm doing some LaTeX), then I'll always compile the same way (e.g., DVI to PS to PDF). I have to be consistent with myself. If ever I change my mind and decide to produce the pdf first (PDFLATEX then PDF2PS), then it's not difficult to do an erase before recompiling, to ensure make does not miss my point. What I mean is that my "generic" use of the make tool is not always optimal in every circumstance. When I do something in the context of a specific project, I create a special makefile for that project. That would solve the PS-PDF out-of-synch problem (well in fact it would not solve it, but it would force me to use always the same sequence to compile because only that one would have been encoded in the makefile, thus the problem would be avoided). Anyway, for an everyday use, I think that a generic use of make can provide very useful services. Waiting for reactions... Olivier _______________________________________________ Help-make mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-make
