Larry Jones wrote:
>
> Todd Denniston writes:
> >
> > with (CVS) 1.9 (client/server) ---as shipped with slackware 3.2
> > "cvs diff -ybw" would give me the familiar side by side diff I want but with
> > (CVS) 1.10.8 (client/server) --- compiled by me on slack 7.0 with
> > ./configure;make;make install
> > "cvs diff -ybw" only gives
> > diff: invalid option -- y
> > Usage: cvs diff.....
> >
> > looking at cvs-1.10.8/diff/diff.c it looks as though the y should be processed if
> > the option ever made it there, but by putting a printf in the y case I know it
> > never does.
>
> The following comment in src/diff.c tells the story:
>
> /* CVS 1.9 and similar versions seemed to have pretty weird handling
> of -y and -T. In the cases where it called rcsdiff,
<SNIP>
> mostly to ignore -q. Maybe this should be fixed, but I think it's
> a larger issue than the changes included here. */
>
> > or maybe if I can find time to do it where to start on a fix, BIG maybe?
>
> I'd start by comparing the options lists in src/diff.c, diff/diff.c, and
> the rcsdiff man page. And if you do decide to do it, please change the
> diff_usage message (in src/diff.c) to document all of the options rather
> than deferring to "rcsdiff-options". (And, although I haven't looked, I
> suspect a similar change ought to be made to the Cederqvist manual.)
>
> -Larry Jones
>
> Pitiful. Just pitiful. -- Calvin
Thanks for the pointers, and I love those C&H comments especially when they hit the
context :).
- I have not looked at changing Cederqvist, I hate doing documentation but love to
have it. :)
- I have updated the diff_usage message.
note: I looked through and tried to remove any option that did not show up in the
getopt_long call or in longopts[], but I could have missed one.
- from what I remember I think people on the list have suggested changing the
following line
"(The most popular is -c for context diffs but there are many more).\n"
to indicate that to generate patches the -u option was there i.e.
"(The most popular is -c for context diffs and -u for patches).\n"
but as I could not find you or Greg indicating as such I did not change it.
-I have done some local testing of various options and could not find that I had
broken anything so this
should be good.(crosses fingers and grabs the asbestos)
--- diff.c 2000/04/01 00:27:55 1.1
+++ diff.c 2000/04/01 00:39:24
@@ -64,6 +64,7 @@
/* FIXME: should be documenting all the options here. They don't
perfectly match rcsdiff options (for example, we always support
--ifdef and --context, but rcsdiff only does if diff does). */
+/*should this FIXME comment go away now? -- Todd*/
static const char *const diff_usage[] =
{
"Usage: %s %s [-lNR] [rcsdiff-options]\n",
@@ -76,8 +77,67 @@
"\t-r rev1\tDiff revision for rev1 against working file.\n",
"\t-r rev2\tDiff rev1/date1 against rev2.\n",
"\t--ifdef=arg\tOutput diffs in ifdef format.\n",
- "(consult the documentation for your diff program for rcsdiff-options.\n",
- "The most popular is -c for context diffs but there are many more).\n",
+ " rcsdiff-options are as follows:\n",
+
+"\t-i --ignore-case Consider upper- and lower-case to be the same.\n",
+"\t-w --ignore-all-space Ignore all white space.\n",
+"\t-b --ignore-space-change Ignore changes in the amount of white space.\n",
+"\t-B --ignore-blank-lines Ignore changes whose lines are all blank.\n",
+"\t-I RE --ignore-matching-lines=RE Ignore changes whose lines all match RE.\n",
+"\t--binary Read and write data in binary mode.\n",/*it is in the longopts
+ i do not know if it
+ works*/
+"\t-a --text Treat all files as text.\n\n",
+"\t-c -C NUM --context[=NUM] Output NUM (default 2) lines of copied context.\n",
+"\t-u -U NUM --unified[=NUM] Output NUM (default 2) lines of unified context.\n",
+"\t -NUM Use NUM context lines.\n",
+"\t -L LABEL --label LABEL Use LABEL instead of file name.\n",
+"\t -p --show-c-function Show which C function each change is in.\n",
+"\t -F RE --show-function-line=RE Show the most recent line matching RE.\n",
+"\t-e --ed Output an ed script.\n",
+"\t-n --rcs Output an RCS format diff.\n",
+"\t-y --side-by-side Output in two columns.\n",
+"\t --width=NUM Output at most NUM (default 130) characters per line.\n",
+"\t --left-column Output only the left column of common lines.\n",
+"\t --suppress-common-lines Do not output common lines.\n",
+"\t-DNAME --ifdef=NAME Output merged file to show `#ifdef NAME' diffs.\n",
+"\t--GTYPE-group-format=GFMT Similar, but format GTYPE input groups with GFMT.\n",
+"\t--line-format=LFMT Similar, but format all input lines with LFMT.\n",
+"\t--LTYPE-line-format=LFMT Similar, but format LTYPE input lines with LFMT.\n",
+"\t LTYPE is `old', `new', or `unchanged'. GTYPE is LTYPE or `changed'.\n",
+"\t GFMT may contain:\n",
+"\t %< lines from FILE1\n",
+"\t %> lines from FILE2\n",
+"\t %= lines common to FILE1 and FILE2\n",
+"\t %[-][WIDTH][.[PREC]]{doxX}LETTER printf-style spec for LETTER\n",
+"\t LETTERs are as follows for new group, lower case for old group:\n",
+"\t F first line number\n",
+"\t L last line number\n",
+"\t N number of lines = L-F+1\n",
+"\t E F-1\n",
+"\t M L+1\n",
+"\t LFMT may contain:\n",
+"\t %L contents of line\n",
+"\t %l contents of line, excluding any trailing newline\n",
+"\t %[-][WIDTH][.[PREC]]{doxX}n printf-style spec for input line number\n",
+"\t Either GFMT or LFMT may contain:\n",
+"\t %% %\n",
+"\t %c'C' the single character C\n",
+"\t %c'\\OOO' the character with octal code OOO\n\n",
+"\t-l --paginate Pass the output through `pr' to paginate it.\n",
+"\t-t --expand-tabs Expand tabs to spaces in output.\n",
+/* "\t-T --initial-tab Make tabs line up by prepending a tab.\n\n", */
+ /* there is a conflict between -T in diff(the above) and rcsdiff
+ (Preserve the modification time on the RCS file...)
+ we could do either but as it is a conflict we do neither. --Todd*/
+"\t-N --new-file Treat absent files as empty.\n",
+"\t-s --report-identical-files Report when two files are the same.\n",
+"\t--horizon-lines=NUM Keep NUM lines of the common prefix and suffix.\n",
+"\t-d --minimal Try hard to find a smaller set of changes.\n",
+"\t-H --speed-large-files Assume large files and many scattered small changes.\n\n",
+
+ "\n",
+ "(The most popular is -c for context diffs but there are many more).\n",
"(Specify the --help global option for a list of other help options)\n",
NULL
};
@@ -138,7 +198,7 @@
{"report-identical-files", 0, 0, 's'},
{"expand-tabs", 0, 0, 't'},
{"ignore-all-space", 0, 0, 'w'},
- {"side-by-side", 0, 0, 147},
+ {"side-by-side", 0, 0, 'y'},
{"unified", 2, 0, 146},
{"left-column", 0, 0, 129},
{"suppress-common-lines", 0, 0, 130},
@@ -185,6 +245,9 @@
anything.) For the purposes of producing output, CVS diff appears
mostly to ignore -q. Maybe this should be fixed, but I think it's
a larger issue than the changes included here. */
+/* the -y option was not used by rcsdiff, only passed to diff.
+ -y: have diff output side by side differences.
+ the other options are noted correctly. -- Todd*/
static void strcat_and_allocate PROTO ((char **, size_t *, const char *));
@@ -246,7 +309,7 @@
optind = 0;
while ((c = getopt_long (argc, argv,
- "+abcdefhilnpstuw0123456789BHNRC:D:F:I:L:U:V:W:k:r:",
+ "+abcdefhilnpstuyw0123456789BHNRC:D:F:I:L:U:V:W:k:r:",
longopts, &option_index)) != -1)
{
switch (c)
@@ -255,7 +318,7 @@
case 'h': case 'i': case 'n': case 'p': case 's': case 't':
case 'u': case 'w': case '0': case '1': case '2':
case '3': case '4': case '5': case '6': case '7': case '8':
- case '9': case 'B': case 'H':
+ case '9': case 'B': case 'H': case 'y':
(void) sprintf (tmp, " -%c", (char) c);
strcat_and_allocate (&opts, &opts_allocated, tmp);
break;
@@ -283,7 +346,7 @@
case 129: case 130: case 132: case 133: case 134:
case 135: case 136: case 137: case 138: case 139: case 140:
case 141: case 142: case 143: case 144: case 145: case 146:
- case 147: case 148:
+ case 148:
strcat_and_allocate (&opts, &opts_allocated, " --");
strcat_and_allocate (&opts, &opts_allocated,
longopts[option_index].name);
The opinions expressed here are not sanctioned by and do not necessarily
represent those of my employer.
