Make !pattern in .gitattributes non-fatal
PATCH 1/1 Make !pattern in .gitattributes non-fatal
Before 82dce99 (attr: more matching optimizations from .gitignore, 2012-10-15), .gitattributes did not have any special treatment of a leading '!'. The docs, however, always said
The rules how the pattern matches paths are the same as in .gitignore files; see linkgit:gitignore[5].
By those rules, leading '!' means pattern negation. So 82dce99 correctly determined that this kind of line makes no sense and should be disallowed.
However, users who actually had a rule for files starting with a '!' are in a bad position: before 82dce99 '!' matched that literal character, so it is conceivable that users have .gitattributes with such lines in them. After 82dce99 the unescaped version was disallowed in such a way that git outright refuses to run(!) most commands in the presence of such a .gitattributes. It therefore becomes very hard to fix, let alone work with, such repositories.
Let's at least allow the users to fix their repos: change the fatal error into a warning.
--- attr.c | 8 +++++--- t/t0003-attributes.sh | 3 ++- 2 files changed, 7 insertions(+), 4 deletions(-)
attr.c +5 -3
@@ -255,9 +255,11 @@ static struct match_attr *parse_attr_line(const char *line, const char *src, &res->u.pat.patternlen, &res->u.pat.flags, &res->u.pat.nowildcardlen); -if (res->u.pat.flags &EXC_FLAG_NEGATIVE ) -die (_ ("Negative patterns are forbidden in git attributes\n" -"Use '\\!' for literal leading exclamation." )); +if (res->u.pat.flags &EXC_FLAG_NEGATIVE ) { +warning (_ ("Negative patterns are ignored in git attributes\n" +"Use '\\!' for literal leading exclamation." )); +return NULL ; + } } res->is_macro = is_macro; res->num_attr = num_attr;
t/t0003-attributes.sh +2 -1
@@ -198,7 +198,8 @@ test_expect_success 'root subdir attribute test' ' test_expect_success 'negative patterns' ' echo "!f test=bar" >.gitattributes && - test_must_fail git check-attr test -- f + git check-attr test -- '"'"'!f'"'"' 2>errors && + test_i18ngrep "Negative patterns are ignored" errors ' test_expect_success 'patterns starting with exclamation' '
Signature · 1 line
1.8.2.rc1.392.gf57d6b8.dirty
PATCH 1/1 attr: always treat the leading exclamation mark as literal
8b1bd02 (Make !pattern in .gitattributes non-fatal - 2013-03-01) raises the fact that even though '!' should have been a negative pattern indicator, people have been misusing for plain '!' and old Git versions accept that. This makes the leading '!' a de facto literal. We can't ever make it negative indicator again because we can't distinguish the literal '!' in existing repos and the negative '!' in future repos.
Acknowledge it in document and code that the leading '!' is forever literal.
---
Perhaps this on top of Thomas' patch? I see no way else that we can change the meaning of '!' again if such patterns are already stored in repositories.
Documentation/gitattributes.txt | 3 ++- attr.c | 8 ++------ dir.c | 7 ++++--- dir.h | 6 +++++- t/t0003-attributes.sh | 5 ++--- 5 files changed, 15 insertions(+), 14 deletions(-)
Documentation/gitattributes.txt +2 -1
@@ -56,7 +56,8 @@ When more than one pattern matches the path, a later line overrides an earlier line. This overriding is done per attribute. The rules how the pattern matches paths are the same as in `.gitignore` files; see linkgit:gitignore[5]. -Unlike `.gitignore`, negative patterns are forbidden. +Unlike `.gitignore`, the leading `!` is treated as a literal +exclamation, not an indicator of negative patterns. When deciding what attributes are assigned to a path, git consults `$GIT_DIR/info/attributes` file (which has the highest
attr.c +2 -6
@@ -254,12 +254,8 @@ static struct match_attr *parse_attr_line(const char *line, const char *src,parse_exclude_pattern (&res->u.pat.pattern, &res->u.pat.patternlen, &res->u.pat.flags, - &res->u.pat.nowildcardlen); -if (res->u.pat.flags &EXC_FLAG_NEGATIVE ) { -warning (_ ("Negative patterns are ignored in git attributes\n" -"Use '\\!' for literal leading exclamation." )); -return NULL ; - } + &res->u.pat.nowildcardlen, +1 ); } res->is_macro = is_macro; res->num_attr = num_attr;
dir.c +4 -3
@@ -311,13 +311,14 @@ static int no_wildcard(const char *string)void parse_exclude_pattern (const char **pattern,int *patternlen,int *flags, -int *nowildcardlen) +int *nowildcardlen, +int gitattributes) {const char *p = *pattern;size_t i, len; *flags =0 ; -if (*p =='!' ) { +if (!gitattributes && *p =='!' ) { *flags |=EXC_FLAG_NEGATIVE ; p++; }
@@ -354,7 +355,7 @@ void add_exclude(const char *string, const char *base,int flags;int nowildcardlen; -parse_exclude_pattern (&string, &patternlen, &flags, &nowildcardlen); +parse_exclude_pattern (&string, &patternlen, &flags, &nowildcardlen,0 );if (flags &EXC_FLAG_MUSTBEDIR ) {char *s; x =xmalloc (sizeof (*x) + patternlen +1 );
dir.h +5 -1
@@ -107,7 +107,11 @@ extern int path_excluded(struct path_exclude_check *, const char *, int namelen,extern int add_excludes_from_file_to_list (const char *fname,const char *base,int baselen,char **buf_p,struct exclude_list *which,int check_index);extern void add_excludes_from_file (struct dir_struct *,const char *fname); -extern void parse_exclude_pattern (const char **string,int *patternlen,int *flags,int *nowildcardlen); +extern void parse_exclude_pattern (const char **string, +int *patternlen, +int *flags, +int *nowildcardlen, +int gitattributes);extern void add_exclude (const char *string,const char *base,int baselen,struct exclude_list *which);extern void free_excludes (struct exclude_list *el);
t/t0003-attributes.sh +2 -3
@@ -197,9 +197,8 @@ test_expect_success 'root subdir attribute test' ' ' test_expect_success 'negative patterns' ' - echo "!f test=bar" >.gitattributes && - git check-attr test -- '"'"'!f'"'"' 2>errors && - test_i18ngrep "Negative patterns are ignored" errors + echo "!f test=foo" >.gitattributes && + attr_check "!f" foo ' test_expect_success 'patterns starting with exclamation' '
Signature · 1 line
1.8.1.2.536.gf441e6d
Quoted 11 lines · Thomas Rast
Before 82dce99 (attr: more matching optimizations from .gitignore, 2012-10-15), .gitattributes did not have any special treatment of a leading '!'. The docs, however, always said
The rules how the pattern matches paths are the same as in
.gitignorefiles; see linkgit:gitignore[5].By those rules, leading '!' means pattern negation. So 82dce99 correctly determined that this kind of line makes no sense and should be disallowed.
Quoted 7 lines
However, users who actually had a rule for files starting with a '!' are in a bad position: before 82dce99 '!' matched that literal character, so it is conceivable that users have .gitattributes with such lines in them. After 82dce99 the unescaped version was disallowed in such a way that git outright refuses to run(!) most commands in the presence of such a .gitattributes. It therefore becomes very hard to fix, let alone work with, such repositories.
Fixing the working tree is easy, but when we read from a history that already records such an entry in an attribute file, it would become somewhat cumbersome. I wouldn't use "very hard to fix" to describe such a case.
But the demotion to warning does make sense; let's do that in v1.8.1.5.
Quoted 21 lines · Junio C Hamano
Quoted 11 lines · Thomas Rast
Before 82dce99 (attr: more matching optimizations from .gitignore, 2012-10-15), .gitattributes did not have any special treatment of a leading '!'. The docs, however, always said
The rules how the pattern matches paths are the same as in
.gitignorefiles; see linkgit:gitignore[5].By those rules, leading '!' means pattern negation. So 82dce99 correctly determined that this kind of line makes no sense and should be disallowed.
Quoted 7 lines
However, users who actually had a rule for files starting with a '!' are in a bad position: before 82dce99 '!' matched that literal character, so it is conceivable that users have .gitattributes with such lines in them. After 82dce99 the unescaped version was disallowed in such a way that git outright refuses to run(!) most commands in the presence of such a .gitattributes. It therefore becomes very hard to fix, let alone work with, such repositories.
Quoted 4 lines
Fixing the working tree is easy, but when we read from a history that already records such an entry in an attribute file, it would become somewhat cumbersome. I wouldn't use "very hard to fix" to describe such a case.
Well, I'm sorry if I hurt any feelings there, but...
~/tmp/badattr(master)$ git show bad:.gitattributes
!bad text
~/tmp/badattr(master)$ ~/g/git-checkout bad # a git without my patch
fatal: Negative patterns are forbidden in git attributes Use '\!' for literal leading exclamation. ~/tmp/badattr(master)$ git status
# On branch master
nothing to commit (use -u to show untracked files)
Notice how it remains on master. I suppose with enough knowledge of the internals I could manage, but after seeing how hard it was to *build* such broken history with a git that dies, I don't really want to try.
Signature · 2 lines
Thomas Rast
trast@{inf,student}.ethz.chQuoted 39 lines · Thomas Rast
Quoted 21 lines · Junio C Hamano
Quoted 11 lines · Thomas Rast
Before 82dce99 (attr: more matching optimizations from .gitignore, 2012-10-15), .gitattributes did not have any special treatment of a leading '!'. The docs, however, always said
The rules how the pattern matches paths are the same as in
.gitignorefiles; see linkgit:gitignore[5].By those rules, leading '!' means pattern negation. So 82dce99 correctly determined that this kind of line makes no sense and should be disallowed.
Quoted 7 lines
However, users who actually had a rule for files starting with a '!' are in a bad position: before 82dce99 '!' matched that literal character, so it is conceivable that users have .gitattributes with such lines in them. After 82dce99 the unescaped version was disallowed in such a way that git outright refuses to run(!) most commands in the presence of such a .gitattributes. It therefore becomes very hard to fix, let alone work with, such repositories.
Quoted 4 lines
Fixing the working tree is easy, but when we read from a history that already records such an entry in an attribute file, it would become somewhat cumbersome. I wouldn't use "very hard to fix" to describe such a case.
Well, I'm sorry if I hurt any feelings there, but...
~/tmp/badattr(master)$ git show bad:.gitattributes
!bad text
~/tmp/badattr(master)$ ~/g/git-checkout bad # a git without my patch
fatal: Negative patterns are forbidden in git attributes Use '\!' for literal leading exclamation. ~/tmp/badattr(master)$ git status
# On branch master
nothing to commit (use -u to show untracked files)
Quoted 2 lines
Notice how it remains on master. I suppose with enough knowledge of the internals I could manage,...
That would have been nicer to have in the log message.
In any case, 1.8.1.5 will go out with this fix (this is v1.8.1 regression IIUC) this afternoon. Thanks for a fix.
Makes sense. Duy?
Quoted 19 lines · Thomas Rast
Before 82dce99 (attr: more matching optimizations from .gitignore, 2012-10-15), .gitattributes did not have any special treatment of a leading '!'. The docs, however, always said
The rules how the pattern matches paths are the same as in
.gitignorefiles; see linkgit:gitignore[5].By those rules, leading '!' means pattern negation. So 82dce99 correctly determined that this kind of line makes no sense and should be disallowed.
However, users who actually had a rule for files starting with a '!' are in a bad position: before 82dce99 '!' matched that literal character, so it is conceivable that users have .gitattributes with such lines in them. After 82dce99 the unescaped version was disallowed in such a way that git outright refuses to run(!) most commands in the presence of such a .gitattributes. It therefore becomes very hard to fix, let alone work with, such repositories.
Quoted 2 lines
Let's at least allow the users to fix their repos: change the fatal error into a warning.
I agree we should not break existing .gitattributes. So yes, die()ing here sounds bad. But..
attr.c static struct match_attr *parse_attr_line(const char *line, const char *src, · line 262
Quoted 12 lines
attr.c +5
-3@@ -255,9 +255,11 @@ static struct match_attr *parse_attr_line(const char *line, const char *src, &res->u.pat.patternlen, &res->u.pat.flags, &res->u.pat.nowildcardlen); -if (res->u.pat.flags &EXC_FLAG_NEGATIVE ) -die (_ ("Negative patterns are forbidden in git attributes\n" -"Use '\\!' for literal leading exclamation." )); +if (res->u.pat.flags &EXC_FLAG_NEGATIVE ) { +warning (_ ("Negative patterns are ignored in git attributes\n" +"Use '\\!' for literal leading exclamation." )); +return NULL ; + }
This "return NULL;" means we ignore "!blah" pattern, which is a regression, isn't it? Should we treat '!' as literal here?
Signature · 1 line
Duy
Quoted 2 lines · Duy Nguyen
This "return NULL;" means we ignore "!blah" pattern, which is a regression, isn't it? Should we treat '!' as literal here?
Probably not. Can you point to a project everybody has heard of that keeps track of a path that begins with an exclamation point?
With clarification to the documentation that says "you cannot use ! to negate" and your "die on such an entry", we have been going in the direction that forbids the use of such an entry, and making it mean literal exclamation point is going sideways in the middle of the road.
Besides, if you want to match a path that begins with an exclam, you can always say "[!]", no?