On Fri, Oct 18, 2019 at 11:50:54AM +0200, Phil Sutter wrote: > Hi Pablo, > > On Fri, Oct 18, 2019 at 10:11:24AM +0200, Pablo Neira Ayuso wrote: > > On Fri, Oct 18, 2019 at 12:48:31AM +0200, Phil Sutter wrote: > > > The same piece of code appears three times, introduce a function to take > > > care of tokenizing and error reporting. > > > > > > Pass buffer pointer via reference so it can be updated to point to after > > > the counters (if found). > > > > > > While being at it, drop pointless casting when passing pcnt/bcnt to > > > add_argv(). > > > > > > Signed-off-by: Phil Sutter <phil@xxxxxx> > > > > Acked-by: Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> > > > > If you get to consolidate more common code between xml and native > > parsers, probably you can add a xtables-restore.c file to store all > > these functions are common, just an idea for the future. > > I get the point, but we have xtables-restore.c already. Though it > contains *tables-nft-restore code. I would add to xshared.c until we > decide it's large enough to split (currently ~750 lines). AFAICT, this > is the only source file included in both xtables-*-multi binaries. Other > than that, I could extend libxtables to really share code but it's > probably not worth it. OK. No libxtables for this, I'd suggest. This is a library and we would need to deal with libversion if you expose this API, xshared.c is fine, I was just wondering if splitting it to encapsulate all shared parsing code in another file could be useful at some point.