Thread: Add const to values and nulls arguments
There are a lot of Datum *values, bool *nulls argument pairs that should really be const. The 0001 patch makes those changes. Some of these hunks depend on each other. The 0002 patch, which I'm not proposing to commit at this time, makes similar changes but in a way that breaks the table and index AM APIs. So I'm just including that here in case anyone wonders, why didn't you touch those. And also maybe if we ever change that API incompatibly we could throw this one in then.
Attachment
Hi, > There are a lot of Datum *values, bool *nulls argument pairs that should > really be const. The 0001 patch makes those changes. Some of these > hunks depend on each other. > > The 0002 patch, which I'm not proposing to commit at this time, makes > similar changes but in a way that breaks the table and index AM APIs. > So I'm just including that here in case anyone wonders, why didn't you > touch those. And also maybe if we ever change that API incompatibly we > could throw this one in then. 0001 looks good to me and passes the tests in several environments, not surprisingly. Let's merge it. -- Best regards, Aleksander Alekseev
On 06.10.23 16:51, Aleksander Alekseev wrote: >> There are a lot of Datum *values, bool *nulls argument pairs that should >> really be const. The 0001 patch makes those changes. Some of these >> hunks depend on each other. >> >> The 0002 patch, which I'm not proposing to commit at this time, makes >> similar changes but in a way that breaks the table and index AM APIs. >> So I'm just including that here in case anyone wonders, why didn't you >> touch those. And also maybe if we ever change that API incompatibly we >> could throw this one in then. > > 0001 looks good to me and passes the tests in several environments, > not surprisingly. Let's merge it. done
Hi, > >> The 0002 patch, which I'm not proposing to commit at this time, makes > >> similar changes but in a way that breaks the table and index AM APIs. > >> So I'm just including that here in case anyone wonders, why didn't you > >> touch those. And also maybe if we ever change that API incompatibly we > >> could throw this one in then. > > > > 0001 looks good to me and passes the tests in several environments, > > not surprisingly. Let's merge it. > > done Great. FWIW changing the index AM API in this particular aspect doesn't strike me as such a terrible idea. -- Best regards, Aleksander Alekseev