I'm pretty sure the logic checks out. I've just run it on the command line to check again. found_group increments whenever the group is in any of the databases. If it's in _none_ of the databases (ie you've !found the group) then found_all is permanently set to 1 so that an error will be returned, and it moves onto searching for the next group. If a group is found then found_all isn't touched, and it moves onto searching for the next group. I think I set found_all to 1 originally because I was going to do something slightly different but I'll initialise it as 0 and rename it to not_found_all when I next update the patch. On 12 July 2013 20:54, Simon Gomizelj wrote:

On Fri, Jul 12, 2013 at 04:56:32PM +0100, awhetter.2011@my.bristol.ac.uk wrote:

...

From: Ashley Whetter

Non-zero is now returned if a group is searched for that doesn't exist

Signed-off-by: Ashley Whetter --- src/pacman/sync.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/pacman/sync.c b/src/pacman/sync.c index fc1314b..35a5c59 100644 --- a/src/pacman/sync.c +++ b/src/pacman/sync.c @@ -367,6 +367,7 @@ static int sync_search(alpm_list_t *syncs, alpm_list_t *targets) static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) { alpm_list_t *i, *j, *k, *s = NULL; + int found_group = 0, found_all = 1;

if(targets) { for(i = targets; i; i = alpm_list_next(i)) { @@ -376,6 +377,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_group_t *grp = alpm_db_get_group(db, grpname);

if(grp) { + found_group++; /* get names of packages in group */ for(k = grp->packages; k; k = alpm_list_next(k)) { if(!config->quiet) { @@ -387,6 +389,11 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) } } } + + if (!found_group) { + found_all = 0; + } + found_group = 0; }

Unless I misunderstood something, I don't think this is the logic you described. You're counting each time a group is found in a given db, and then checking that the count is non-zero. Thus you'll only trigger a return code of 1 if _none_ of your searches result in a match.

It should return 1 if any one search fails.

...

} else { for(i = syncs; i; i = alpm_list_next(i)) { @@ -394,6 +401,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets)

for(j = alpm_db_get_groupcache(db); j; j = alpm_list_next(j)) { alpm_group_t *grp = j->data; + found_all = 1;

found_all is already set to one. This else block is completely independent.

...

- return 0; + return !found_all;

The negation here seems arbitrary and needless. You're explicitly setting found_all to either 1 or 0 in the first place.

On 14/07/13 02:01, awhetter.2011@my.bristol.ac.uk wrote:

From: Ashley Whetter

Non-zero is now returned if a group is searched for that doesn't exist Fixes FS#36097

Signed-off-by: Ashley Whetter --- In this version I've renamed the variables as Simon suggested. Also I noticed that when no targets were specified the behaviour of -Ss wasn't mirrored and a success would always be returned. So I've have to reinitialise 'found' if there are no targets, which isn't ideal but the only other decent options I could think of was to either name 'found' as 'not_found' and flip the 0s and 1s accordingly, or to check the length of alpm_db_get_groupcache(db) by taking the 'j =' initiasation out of the for loop, which seemed worse. Also the way I'm returning 'found' is inconsistent with how it's done in sync_search, so I feel like sync_search should be changed as well because '!found' is more readable than '(found == 0)'. src/pacman/sync.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/pacman/sync.c b/src/pacman/sync.c index fc1314b..313c3cd 100644 --- a/src/pacman/sync.c +++ b/src/pacman/sync.c @@ -367,6 +367,7 @@ static int sync_search(alpm_list_t *syncs, alpm_list_t *targets) static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) { alpm_list_t *i, *j, *k, *s = NULL; + int found_in_db = 0, found = 1;

I'm going to ask for a change here, which is mostly just a variable renaming... but requires other minor adjustment. The "found" variable is really just tracking our return value. So name it that. Also "found_in_db" makes me concerned that things can be found outside a db! So: found_in_db -> found found -> ret

if(targets) {

found_in_db (-> found) is only used in this scope so define it here.

for(i = targets; i; i = alpm_list_next(i)) { @@ -376,6 +377,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_group_t *grp = alpm_db_get_group(db, grpname);

if(grp) { + found_in_db++; /* get names of packages in group */ for(k = grp->packages; k; k = alpm_list_next(k)) { if(!config->quiet) { @@ -387,13 +389,19 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) } } } + if (!found_in_db) { + found = 0; + } + found_in_db = 0;

Set this at the start of the loop.

} } else { + found = 0; for(i = syncs; i; i = alpm_list_next(i)) { alpm_db_t *db = i->data;

for(j = alpm_db_get_groupcache(db); j; j = alpm_list_next(j)) { alpm_group_t *grp = j->data; + found = 1;

if(level > 1) { for(k = grp->packages; k; k = alpm_list_next(k)) { @@ -412,7 +420,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_list_free(s); }

- return 0; + return !found; }

static int sync_info(alpm_list_t *syncs, alpm_list_t *targets)

On 15/07/13 17:42, Ashley Whetter wrote:

On 15 July 2013 06:19, Allan McRae wrote:

...

On 14/07/13 02:01, awhetter.2011@my.bristol.ac.uk wrote:

...

From: Ashley Whetter

Non-zero is now returned if a group is searched for that doesn't exist Fixes FS#36097

Signed-off-by: Ashley Whetter --- In this version I've renamed the variables as Simon suggested. Also I noticed that when no targets were specified the behaviour of -Ss wasn't mirrored and a success would always be returned. So I've have to reinitialise 'found' if there are no targets, which isn't ideal but the only other decent options I could think of was to either name 'found' as 'not_found' and flip the 0s and 1s accordingly, or to check the length of alpm_db_get_groupcache(db) by taking the 'j =' initiasation out of the for loop, which seemed worse. Also the way I'm returning 'found' is inconsistent with how it's done in sync_search, so I feel like sync_search should be changed as well because '!found' is more readable than '(found == 0)'.

What about this issue with returning '!found' or '(found == 0)'? Should I change how I'm doing it, or how sync_search and any other similar functions are doing it? Or neither?

I did not answer because following my suggestions that variable will be called ret and store the return value.

...

...

src/pacman/sync.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/src/pacman/sync.c b/src/pacman/sync.c index fc1314b..313c3cd 100644 --- a/src/pacman/sync.c +++ b/src/pacman/sync.c @@ -367,6 +367,7 @@ static int sync_search(alpm_list_t *syncs, alpm_list_t *targets) static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) { alpm_list_t *i, *j, *k, *s = NULL; + int found_in_db = 0, found = 1;

I'm going to ask for a change here, which is mostly just a variable renaming... but requires other minor adjustment. The "found" variable is really just tracking our return value. So name it that. Also "found_in_db" makes me concerned that things can be found outside a db! So:

found_in_db -> found found -> ret

...

if(targets) {

found_in_db (-> found) is only used in this scope so define it here.

I hadn't thought about that. Thanks for the feedbck. I'll get those changed.

...

...

for(i = targets; i; i = alpm_list_next(i)) { @@ -376,6 +377,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_group_t *grp = alpm_db_get_group(db, grpname);

if(grp) { + found_in_db++; /* get names of packages in group */ for(k = grp->packages; k; k = alpm_list_next(k)) { if(!config->quiet) { @@ -387,13 +389,19 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) } } } + if (!found_in_db) { + found = 0; + } + found_in_db = 0;

Set this at the start of the loop.

...

} } else { + found = 0; for(i = syncs; i; i = alpm_list_next(i)) { alpm_db_t *db = i->data;

for(j = alpm_db_get_groupcache(db); j; j = alpm_list_next(j)) { alpm_group_t *grp = j->data; + found = 1;

if(level > 1) { for(k = grp->packages; k; k = alpm_list_next(k)) { @@ -412,7 +420,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_list_free(s); }

- return 0; + return !found; }

static int sync_info(alpm_list_t *syncs, alpm_list_t *targets)

On 16/07/13 19:18, awhetter.2011@my.bristol.ac.uk wrote:

From: Ashley Whetter

Non-zero is now returned if a group is searched for that doesn't exist Fixes FS#36097

Please us proper punctuation in commit messages. That means fullstops at the end of sentences. I will fix when committing.

Signed-off-by: Ashley Whetter --- The only changes in this version are the one's recommended by Allan. 'found' is now called 'ret' and 'found_in_db' is now called 'found'. 'found_in_db' (now 'found') is initialised in the 'if (targets)' condition. src/pacman/sync.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/src/pacman/sync.c b/src/pacman/sync.c index fc1314b..e211f7b 100644 --- a/src/pacman/sync.c +++ b/src/pacman/sync.c @@ -367,8 +367,10 @@ static int sync_search(alpm_list_t *syncs, alpm_list_t *targets) static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) { alpm_list_t *i, *j, *k, *s = NULL; + int ret = 0;

if(targets) { + int found = 0; for(i = targets; i; i = alpm_list_next(i)) { const char *grpname = i->data; for(j = syncs; j; j = alpm_list_next(j)) { @@ -376,6 +378,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_group_t *grp = alpm_db_get_group(db, grpname);

if(grp) { + found++; /* get names of packages in group */ for(k = grp->packages; k; k = alpm_list_next(k)) { if(!config->quiet) { @@ -387,13 +390,19 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) } } } + if (!found) { + ret = 1; + } + found = 0;

The only change I am making is to move this to the start of the loop. I'd much rather it is set there than relying on it being set at the end. Then if a "continue" ever gets put in the loop, it will still work.

} } else { + ret = 1; for(i = syncs; i; i = alpm_list_next(i)) { alpm_db_t *db = i->data;

for(j = alpm_db_get_groupcache(db); j; j = alpm_list_next(j)) { alpm_group_t *grp = j->data; + ret = 0;

if(level > 1) { for(k = grp->packages; k; k = alpm_list_next(k)) { @@ -412,7 +421,7 @@ static int sync_group(int level, alpm_list_t *syncs, alpm_list_t *targets) alpm_list_free(s); }

- return 0; + return ret; }

static int sync_info(alpm_list_t *syncs, alpm_list_t *targets)