Re: [pgsql-hackers-win32] Tablespaces
От | Lawrence E. Smithmier, Jr. |
---|---|
Тема | Re: [pgsql-hackers-win32] Tablespaces |
Дата | |
Msg-id | 1078508484.4048bbc4c4403@mail.cbtree.org обсуждение исходный текст |
Список | pgsql-hackers |
Quoting Bruce Momjian <pgman@candle.pha.pa.us>: > Zeugswetter Andreas SB SD wrote: > > My idea for platforms that don't support symlinks would be to simply > create > > a tblspaceoid directory inplace instead of the symlink (maybe throw a > warning). > > My feeling is, that using the same syntax on such platforms is important, > > but actual distribution is not (since they will most likely be small > systems). > > Nice idea --- so you can create tablespaces, but can't specify a > different location for it --- makes sense. If this is the way it is handled then I don't guess I have any objections. You would end up with the files for a single database spread out, possibly over multiple disks. I think I would still prefer the option of using an XML file to this, but I can also see the need to limit the platform specific code. I guess if I feel strongly enough about it, through the magic of Open Source I can patch it to work the way I want. If the way Junctions work on NTFS becomes an issue after the port, it won't be a problem, it will be an opportunity. Quoting Bruce Momjian <pgman@candle.pha.pa.us>: > The use of symlinks on Unix has just too many advantages to use a > another sub-optimal solution on that platform. And creating separate solutions for different platforms has too many disadvantages to warrent a compile time option. Ok, implement this and move on to porting so that I can use it in the world I work in. :-) Seriously, thanks for doing all you guys do. I will be activly evangelizing as soon as I can get a version working standalone where I am. Lawrence E. Smithmier, Jr. MCP, MCAD (919) 522-9738 Larry@Smithmier.com
В списке pgsql-hackers по дате отправления: