Message ID | 20230503184849.1809304-1-calvinwan@google.com (mailing list archive) |
---|---|
Headers | show |
Series | strbuf cleanups | expand |
Calvin Wan <calvinwan@google.com> writes: > Strbuf is a widely used basic structure that should only interact with > other primitives in strbuf.[ch]. Over time certain functions inside of > strbuf.[ch] have been added to interact with higher level objects and > functions. This series cleans up some of those higher level interactions > by moving the offending functions to the files they interact with and > adding documentation to strbuf.h. With the goal of eventually being able > to stand up strbuf as a libary, this series also removes the use of > environment variables from strbuf. This round hasn't seen any comments (mine does not count ;-). The 7/7-only v3 still says "environment variable" to refer to a global variable, so I am not sure where we stand. Is this back-burnered?
On Wed, May 3, 2023 at 11:48 AM Calvin Wan <calvinwan@google.com> wrote: > > Strbuf is a widely used basic structure that should only interact with > other primitives in strbuf.[ch]. "should"? I agree with moving things in this direction, but the wording here suggests there's a pre-existing directive or guideline; but if so, I'm unfamiliar with it. Maybe this should be reworded to "This series modifies strbuf to focus on string manipulation with minimal other dependencies"? > Over time certain functions inside of > strbuf.[ch] have been added to interact with higher level objects and > functions. This series cleans up some of those higher level interactions > by moving the offending functions to the files they interact with and > adding documentation to strbuf.h. With the goal of eventually being able > to stand up strbuf as a libary, this series also removes the use of > environment variables from strbuf. As Junio pointed out, "environment variables" is misleading; see below. > Range-diff against v1: > -: ---------- > 1: e0dd3f5295 strbuf: clarify API boundary I read this separately; this patch looks good to me. > 1: 283771c088 ! 2: ec1ea6ae4f abspath: move related functions to abspath > @@ Commit message > abspath: move related functions to abspath > > Move abspath-related functions from strbuf.[ch] to abspath.[ch] since > - they do not belong in a low-level library. > + paths are not primitive objects and therefore strbuf should not interact > + with them. This applies to patches 4 & 5 as well: Would this perhaps be better worded as: Move abspath-related functions from strbuf.[ch] to abspath.[ch] so that strbuf is focused on string manipulation routines with minimal dependencies. ? > ## abspath.c ## > @@ abspath.c: char *prefix_filename_except_for_dash(const char *pfx, const char *arg) > 2: 58f78b8ae0 = 3: 2d74561b91 credential-store: move related functions to credential-store file > 3: 88ab90c079 ! 4: 30b5e635cb object-name: move related functions to object-name > @@ Commit message > object-name: move related functions to object-name > > Move object-name-related functions from strbuf.[ch] to object-name.[ch] > - since they do not belong in a low-level library. > + since paths are not a primitive object that strbuf should directly > + interact with. > > ## object-name.c ## > @@ object-name.c: static void find_abbrev_len_packed(struct min_abbrev_data *mad) > 4: 30b7de5a81 ! 5: 6905618470 path: move related function to path > @@ Metadata > ## Commit message ## > path: move related function to path > > - Move path-related function from strbuf.[ch] to path.[ch] since it does > - not belong in a low-level library. > + Move path-related function from strbuf.[ch] to path.[ch] since path is > + not a primitive object and therefore strbuf should not directly interact > + with it. > > ## path.c ## > @@ path.c: int normalize_path_copy(char *dst, const char *src) > 5: 7b6d6353de ! 6: caf3482bf7 strbuf: clarify dependency > @@ Metadata > ## Commit message ## > strbuf: clarify dependency > > + refs.h was once needed but is no longer so as of 6bab74e7fb8 ("strbuf: > + move strbuf_branchname to sha1_name.c", 2010-11-06). strbuf.h was > + included thru refs.h, so removing refs.h requires strbuf.h to be added > + back. > + Thanks. > 6: ffacd1cbe5 ! 7: a7f23488f8 strbuf: remove enviroment variables > @@ Metadata > Author: Calvin Wan <calvinwan@google.com> > > ## Commit message ## > - strbuf: remove enviroment variables > + strbuf: remove environment variables > > - As a lower level library, strbuf should not directly access environment > - variables within its functions. Therefore, add an additional variable to > - function signatures for functions that use an environment variable and > - refactor callers to pass in the environment variable. > + As a library that only interacts with other primitives, strbuf should > + not directly access environment variables within its > + functions. Therefore, add an additional variable to function signatures > + for functions that use an environment variable and refactor callers to > + pass in the environment variable. "environment variable" makes one think not of "variable from environment.c [that happens to be a global]" but of https://en.wikipedia.org/wiki/Environment_variable. The fact that the variable is defined in environment.c isn't even relevant here, though, while the fact that it is a global is relevant. I'd just s/environment variable/global variable/ in all 4 places (er, 5 if you include the cover letter). Your v3 7/7 has an important correction, but you didn't send out a range diff for v3 so I'm putting my comments on your v2 range-diff. :-) > + /** > * Strip whitespace from a buffer. The second parameter controls if > - * comments are considered contents to be removed or not. > +- * comments are considered contents to be removed or not. > ++ * comments are considered contents to be removed or not. The third parameter > ++ * is the comment character that determines whether a line is a comment or not. > */ Thanks. This series fixes two of my comments on v1, and the other was just "here's another cleanup that could be added" -- it's fine to punt on. I only had a couple minor comments on this v2/v3 that should be easy to fix up, all included in this response to the cover letter.
Elijah Newren wrote: > On Wed, May 3, 2023 at 11:48 AM Calvin Wan <calvinwan@google.com> wrote: > > > > Strbuf is a widely used basic structure that should only interact with > > other primitives in strbuf.[ch]. > > "should"? I agree with moving things in this direction, but the > wording here suggests there's a pre-existing directive or guideline; No, it doesn't. "You should see a doctor about that cough" is not a directive, it's merely an advice. > Maybe this should be reworded to "This series modifies strbuf to focus on > string manipulation with minimal other dependencies"? And ironoically you use the word "should" to provide your own advice on how to avoid using "should" as an advice. I agree it would be ideal if strbuf interacted only with other primitives in strbuf.[ch], so I agree "should" is a perfectly fine word to describe such a normative statement.
I intend to reroll it, as well as rewording the commit patches 4 and 5 as per Elijah's suggestions. I didn't originally realize you were referring to s/environment variable/global variable as the misnomer (since the dependency I was removing was from environment.h) On Fri, May 5, 2023 at 3:33 PM Junio C Hamano <gitster@pobox.com> wrote: > > Calvin Wan <calvinwan@google.com> writes: > > > Strbuf is a widely used basic structure that should only interact with > > other primitives in strbuf.[ch]. Over time certain functions inside of > > strbuf.[ch] have been added to interact with higher level objects and > > functions. This series cleans up some of those higher level interactions > > by moving the offending functions to the files they interact with and > > adding documentation to strbuf.h. With the goal of eventually being able > > to stand up strbuf as a libary, this series also removes the use of > > environment variables from strbuf. > > This round hasn't seen any comments (mine does not count ;-). The > 7/7-only v3 still says "environment variable" to refer to a global > variable, so I am not sure where we stand. Is this back-burnered? >