Hi Bruno, > Le 30 sept. 2018 à 01:09, Bruno Haible <br...@clisp.org> a écrit : > > Hi Akim, > > How about the following patches: > > * In timevar.c: Include timevar.h first. This is a Gnulib best practice, which > has the benefit of verifying that the header file is self-contained. > > * In timevar.def: > - The DEFTIMEVAR invocation must NOT be followed by a semicolon, otherwise > you'll get a syntax error in the enum definition. > - Enum values are not positive integers. The first enum value is 0. > - What is the difference between a "character string" and a "string"? > - The statement that the timing table is printed in the given order > contradicts the comment in timevar.h. > > * In timevar.h: > - Better talk about the program or the application, not the « compiler".
All this is good, of course. > - It is pointless to omit the parameter names from the function declarations > if the comments talk about TIMEVAR, ELAPSED, FP, etc. Here, I don’t understand why you commented the argument names. It would be clearer with them uncommented. I should have done that when converting to ANSI C. Also, maybe the extern C. I see several modules have it. > * In the module description: > Are you subscribed to bug-gnulib? Nope. > If not, you should be listed as the > maintainer of the modules, so that people CC you when they have questions > or patches about it. Sure!
