Обсуждение: Win32 version numbers not correct (again, but this one is easy)
Seems to be a neverending story :-) Version numbers don't update properly on postgres.exe when doing make clean/make. It works for all other exes and dlls. I beleive this is just a case of "make clean" not properly removing "win32ver.o". See attached patch. //Magnus <<backend_clean.patch>>
Вложения
"Magnus Hagander" <mha@sollentuna.net> writes: > Version numbers don't update properly on postgres.exe when doing make > clean/make. It works for all other exes and dlls. I beleive this is just > a case of "make clean" not properly removing "win32ver.o". See attached > patch. Hmm, is that the only place we missed it? Might be worth diffing a make distclean'd tree against a virgin one. regards, tom lane
Tom Lane wrote: > "Magnus Hagander" <mha@sollentuna.net> writes: > > Version numbers don't update properly on postgres.exe when doing make > > clean/make. It works for all other exes and dlls. I beleive this is just > > a case of "make clean" not properly removing "win32ver.o". See attached > > patch. > > Hmm, is that the only place we missed it? Might be worth diffing a make > distclean'd tree against a virgin one. I was under the impression we only wanted a new version number with a configure run, however, I can see that "make clean" perhaps should do it too. -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073
Bruce Momjian <pgman@candle.pha.pa.us> writes: > Tom Lane wrote: >> Hmm, is that the only place we missed it? Might be worth diffing a make >> distclean'd tree against a virgin one. > I was under the impression we only wanted a new version number with a > configure run, however, I can see that "make clean" perhaps should do it > too. No, it's not the source code file he's complaining about. regards, tom lane
>> Version numbers don't update properly on postgres.exe when doing make >> clean/make. It works for all other exes and dlls. I beleive >this is just >> a case of "make clean" not properly removing "win32ver.o". >See attached >> patch. > >Hmm, is that the only place we missed it? Might be worth >diffing a make >distclean'd tree against a virgin one. I beleive most of these are intentional - like the flex/bison output, but I'll let you call those to be sure. I grabbed a fresh cvs into "pgsql-clean", did "make distclean" in "pgsql" and ran "diff -cr pgsql pgsql-clean -x CVS". Here is the list I got, filtered from the two or three changes I've done myself (this is with the patch I sent for removal of win32ver.o in src/backend): Only in pgsql/src/backend/bootstrap: bootparse.c Only in pgsql/src/backend/bootstrap: bootscanner.c Only in pgsql/src/backend/bootstrap: bootstrap_tokens.h Only in pgsql/src/backend/parser: gram.c Only in pgsql/src/backend/parser: parse.h Only in pgsql/src/backend/parser: scan.c Only in pgsql/src/backend/utils/misc: guc-file.c Only in pgsql/src/bin/psql: psqlscan.c Only in pgsql/src/bin/psql: sql_help.h Only in pgsql/src/interfaces/ecpg/preproc: pgc.c Only in pgsql/src/interfaces/ecpg/preproc: preproc.c Only in pgsql/src/interfaces/ecpg/preproc: preproc.h Only in pgsql/src/interfaces/libpq: blibpqdll.def Only in pgsql/src/interfaces/libpq: libpq.rc Only in pgsql/src/interfaces/libpq: libpqddll.def Only in pgsql/src/interfaces/libpq: libpqdll.def Only in pgsql/src/pl/plpgsql/src: pl.tab.h Only in pgsql/src/pl/plpgsql/src: pl_gram.c Only in pgsql/src/pl/plpgsql/src: pl_scan.c Only in pgsql/src/timezone: zic.exe ^ Seems the Makefile only removes "zic" not "zic.exe". Should be zic$(X), I beleive. //Magnus
"Magnus Hagander" <mha@sollentuna.net> writes: > Only in pgsql/src/timezone: zic.exe > ^ Seems the Makefile only removes "zic" not "zic.exe". Should be > zic$(X), I beleive. Ah, good catch. The rest are supposed to be there. regards, tom lane
"Magnus Hagander" <mha@sollentuna.net> writes: > Version numbers don't update properly on postgres.exe when doing make > clean/make. It works for all other exes and dlls. I beleive this is just > a case of "make clean" not properly removing "win32ver.o". See attached > patch. Applied, except I used $(WIN32RES) which seems more consistent with what's done elsewhere. regards, tom lane