Hi Jean, Am Do., 2. Juli 2020 um 14:44 Uhr schrieb Jean Delvare <[email protected]>: > Hi Andreas, > > All quilt commands start by sourcing scripts/patchfns, but before doing > so, they check whether this file has already been sourced. Looking at > the code, I can't see any case where the file would have been sourced > already (it is being sourced exactly once in each command-specific > file). Can you remember why you added that check?
I'm not quite sure. It doesn't seem to be because of the --trace option, and neither because of recursive calls (quilt_command). > The only thing that came to my mind was that maybe you wanted to make > it possible to pre-source the file so that individual commands do not > need to do it again. That would make sense from a performance > perspective, however I don't think this can be done at the moment, > because a number of checks in scripts/patchfns are context-dependent > (sourcing of $QUILTRC, addition of default arguments for the command, > discovery of the source tree's root, setting of $QUILT_PATCHES, > $QUILT_SERIES and $SERIES, version check and now series check). So it > would only work if we would split non context-dependent code to a > separate file and pre-source only that file. > > Whether we ever want to do that or not, I'm not sure, as it would > seriously clutter the shell's environment, especially considering the > lack of namespace marker for all functions in this file. > > So I am considering removing the check for double inclusion. Do you > have any objection to me doing that? > > If the check is really needed then I'd rather move it to > scripts/patchfns itself so that it isn't duplicated in every > command-specific file, pretty much as is traditionally done for C > header files. Just try it out, I guess. Thanks, Andreas _______________________________________________ Quilt-dev mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/quilt-dev
