mjhjmtu

I noticed there are 47 blanks lines on top of locale files:

xb@dnxb:/usr/share/X11/locale$ hexdump -Cv locale.alias | head
00000000 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000010 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000020 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 23 |...............#|
00000030 0a 23 09 54 68 69 73 20 66 69 6c 65 20 63 6f 6e |.#.This file con|
00000040 74 61 69 6e 73 20 61 6c 69 61 73 20 6e 61 6d 65 |tains alias name|
00000050 20 6f 66 20 6c 6f 63 61 6c 65 2e 0a 23 09 45 61 | of locale..#.Ea|
00000060 63 68 20 61 6c 69 61 73 20 6e 61 6d 65 20 69 73 |ch alias name is|
00000070 20 64 65 73 63 72 69 62 65 64 20 77 69 74 68 69 | described withi|
00000080 6e 20 6f 6e 65 20 6c 69 6e 65 2e 0a 23 09 54 68 |n one line..#.Th|
00000090 65 20 66 69 72 73 74 20 77 6f 72 64 20 69 73 20 |e first word is |
xb@dnxb:/usr/share/X11/locale$ hexdump -Cv locale.dir | head
00000000 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000010 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000020 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 23 |...............#|
00000030 0a 23 09 54 68 69 73 20 66 69 6c 65 20 63 6f 6e |.#.This file con|
00000040 74 61 69 6e 73 20 6c 6f 63 61 6c 65 20 64 61 74 |tains locale dat|
00000050 61 62 61 73 65 20 66 69 6c 65 20 6e 61 6d 65 73 |abase file names|
00000060 0a 23 09 54 68 65 20 66 69 72 73 74 20 77 6f 72 |.#.The first wor|
00000070 64 20 69 73 20 74 68 65 20 6c 6f 63 61 6c 65 20 |d is the locale |
00000080 64 61 74 61 62 61 73 65 20 66 69 6c 65 20 6e 61 |database file na|
00000090 6d 65 20 61 6e 64 0a 23 09 74 68 65 20 73 65 63 |me and.#.the sec|
xb@dnxb:/usr/share/X11/locale$ hexdump -Cv compose.dir | head 
00000000 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000010 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000020 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 23 |...............#|
00000030 0a 23 09 54 68 69 73 20 66 69 6c 65 20 63 6f 6e |.#.This file con|
00000040 74 61 69 6e 73 20 63 6f 6d 70 6f 73 65 20 74 61 |tains compose ta|
00000050 62 6c 65 20 66 69 6c 65 20 6e 61 6d 65 73 2e 0a |ble file names..|
00000060 23 09 54 68 65 20 66 69 72 73 74 20 77 6f 72 64 |#.The first word|
00000070 20 69 73 20 74 68 65 20 63 6f 6d 70 6f 73 65 20 | is the compose |
00000080 74 61 62 6c 65 20 66 69 6c 65 20 6e 61 6d 65 0a |table file name.|
00000090 23 09 61 6e 64 20 74 68 65 20 73 65 63 6f 6e 64 |#.and the second|
xb@dnxb:/usr/share/X11/locale$ hexdump -Cv C/XLC_LOCALE | head
00000000 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000010 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a |................|
00000020 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 0a 23 |...............#|
00000030 20 20 58 4c 6f 63 61 6c 65 20 44 61 74 61 62 61 | XLocale Databa|
00000040 73 65 20 53 61 6d 70 6c 65 20 66 6f 72 20 43 2e |se Sample for C.|
00000050 0a 23 20 0a 0a 23 20 0a 23 20 09 58 4c 43 5f 46 |.# ..# .# .XLC_F|
00000060 4f 4e 54 53 45 54 20 63 61 74 65 67 6f 72 79 0a |ONTSET category.|
00000070 23 20 0a 58 4c 43 5f 46 4f 4e 54 53 45 54 0a 23 |# .XLC_FONTSET.#|
00000080 20 09 66 73 30 20 63 6c 61 73 73 20 28 37 20 62 | .fs0 class (7 b|
00000090 69 74 20 41 53 43 49 49 29 0a 66 73 30 09 7b 0a |it ASCII).fs0.{.|
xb@dnxb:/usr/share/X11/locale$ 

I probably will not scroll down if I only vi the file and see blank page, I only noticed this if I cat it. I check this spec but seems not mention anything about 47 blank lines. What's the purpose of this 47 blank lines at the beginning of locale files ?

I'm pretty sure its not really there for any reason; it's a side-effect of how those files are built. If you take a look at the source for locale.alias, you'll see they're missing (note the numbers on the left are added line numbers, they're not actually present in the file):

1 XCOMM
2 XCOMM This file contains alias name of locale.
3 XCOMM Each alias name is described within one line.
4 XCOMM The first word is the alias name (simplified locale name)
5 XCOMM the second word is full locale name.
6 XCOMM
7 XCOMM

8 POSIX: C
9 POSIX-UTF2: C

If you look at the Makefile.am that is passed through a fair bit of C preprocessor and sed to generate the file:

locale.alias: locale.alias.pre
 $(AM_V_GEN)$(RAWCPP) $(RAWCPPFLAGS) $(CPP_FILES_FLAGS) < $(srcdir)/locale.alias.pre | $(CPP_SED_MAGIC) > locale.alias.l1
 $(SED) -e '/^[^#][^ ]*:/s/://' -e '/^[^#].*[ ].*:/d' 
 < locale.alias.l1 > locale.alias.l2
 cat locale.alias.l2 locale.alias.l1 > locale.alias

If you look at Makefile.in, CPP_SED_MAGIC is:

CPP_SED_MAGIC = $(SED) -e '/^# *[0-9][0-9]* *.*$$/d' 
 -e '/^#line *[0-9][0-9]* *.*$$/d' 
 -e '/^[ ]*XCOMM$$/s/XCOMM/#/' 
 -e '/^[ ]*XCOMM[^a-zA-Z0-9_]/s/XCOMM/#/' 
 -e '/^[ ]*XHASH/s/XHASH/#/' 
 -e 's,X11_LOCALEDATADIR,$(X11_LOCALEDATADIR),g' 
 -e '/@@$$/s/@@$$/\/'

Which, among other things, replaces all the line-number comments the C preprocessor spits out with blank lines. So that's where they're coming from.