On 03/26/2018 11:29 PM, Jon Maloy wrote: >> Sorry, I don't suggest we should make so significant changes in order to >> make function names better. >> >> If the changes are merged into upstream, it's almost impossible for >> users to be able to easily back port TIPC patches from upstream version >> to lower Linux version. Please remember there are many users who cannot >> always upgrade kernel version again and again. > Not sure what you are suggesting here. That I skip all name changes? That I > skip only the function and file name changes? That I skip the whole series?
Sorry for this confusion. > > First, I believed you agreed that the change from an array structure to an RB > structure is a good thing. It solves two important problems which are not > about the code, but about architecture, regarding memory consumption and > table consistency. To me, this is a necessary and fundamental upgrade in > order to get rid of two old, lingering problems. No, I don't objected to these changes. Instead, they are good things you introduced in this series. > > Assuming that this is not you concern, you should give some thought to what > the code will look like inside name_table.c after introduction of the tree > structure, with or without name changes. No patch made from name_table.c > after the structure change will be applicable on older Linux versions, in > either case. The changes are too fundamental. So we gain nothing by *not* > changing those struct and variable names there, either way. Exactly! This is what I expect. > > Regarding the parts visible outside name_table.c and name_distr.c I have some > more understanding for your concern, because those have impact on other files > (socket.c, node.c etc) > > We could do as follows: > Struct name_table -> struct name_table // I.e. no change. Visible > externally via name_table.h > struct publication -> publication // I.e., no change. Visible > externally and used by many other files. > > struct name_seq -> struct tipc_service // Only visible in name_table.c > anyway > struct sub_seq -> struct service_range // Only visible inside name_table.c > struct publ_info is removed, just as I do in the later commits > > We don't change any externally visible function names or file names, unless > maybe when the functions anyway must be changed because of new/removed > parameters. > This would fulfil my intention to make the code in name_table.c more readable > and up to common Linux standards, which I don't consider it to be now. > Thanks, I agree with you. > Regarding the changed API struct names in tipc.h in patch #2, I regard this > as important. The old names are meaningless to any new user, while the new > names give an immediate intuitive understanding of what they are, and are > consistent with how those are described in documentation and presentations. > And it is anyway a fully compatible and harmless change. Yeah, we can change it. Please go ahead with these changes. Thanks, Ying ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ tipc-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/tipc-discussion
