RE: [Ctags] bug in ctags parsing of arguments?

[Coen E. <Coe...@cm...>]

I found out what's going wrong!
Now I'm pretty sure it is a bug, and I think it's not VMS-specific.
See below.

> > Did you test this? I still get the same error:
> 
> The only thing what I changed is --language-force parameter, 
> and it works fine.
> 
> TEST,069 zay_$ ctags --language-force=c --totals -o tags "-L" 
> sources.txt
> 3 files, 848 lines (21 kB) scanned in 0.0 seconds (538 kB/s)
> 55 tags added to tag file
> 55 tags sorted in 0.01 seconds
> 
> My ctags version is:
> 
> TEST,069 zay_$ ctags --version
> Exuberant Ctags 5.0, Copyright (C) 1996-2001 Darren Hiebert
>   Addresses: <da...@hi...>, http://ctags.sourceforge.net
>   Optional compiled features: +vms, +internal-sort
> 
> the very same from www.polarfox.com .


I thought of something else: I downloaded the W95 version of ctags 5.0. That
worked OK with "--totals".
I guess it must be my VMS environment then.   :-(

OK, I rolled op my sleeves, downloaded
http://www.polarfox.com/ctags/files/source/vms/ctags-50-src-vms.zip (as it
turns out, about 2 hours before Zoltan released a 5.01 version  ;-D  ),
compiled a debug-version and fired up the old debugger.

(BTW, I got TONS of warnings from my compiler. I switched all errorchecking
off to get ctags.exe built. I can mail them if anyone's interested, but a
lot were of the caliber "In this statement, the enum variable
"inMultilineString" is used in an arithmetic operation", i.e. not very
serious)

Using "monitor ParsedLeadingOptions" and careful stepping, I determined that
"ParsedLeadingOptions" was initialized to FALSE, but set to "TRUE" BEFORE
the command line arguments are actually parsed. I marked the spot with XXX
below:

$ ctags --verbose --totals
Installing parsers: Asm, Asp, Awk, BETA, C, C++, Cobol, Eiffel, Fortran,
Java, Lisp, Make, Pascal, Perl, PHP, Python, Ruby, Scheme, Sh, Tcl, Vim
Setting option defaults
    Setting default header extensions: h, H, hh, hpp, hxx, h++, inc, def
    Installing default language mappings:
    Asm: .asm .s .S
    Asp: .asp .asa
    Awk: .awk .gawk .mawk
    BETA: .bet
    C: .c
    C++: .c++ .cc .cp .cpp .cxx .h .h++ .hh .hp .hpp .hxx
    Cobol: .cob .COB
    Eiffel: .e
    Fortran: .f .for .ftn .f77 .f90 .f95
    Java: .java
    Lisp: .cl .clisp .el .l .lisp .lsp .ml
    Make: ([Mm]akefile) .mak
    Pascal: .p .pas
    Perl: .pl .pm .perl
    PHP: .php .php3 .phtml
    Python: .py .python
    Ruby: .rb
    Scheme: .SCM .SM .sch .scheme .scm .sm
    Sh: .sh .SH .bsh .bash .ksh .zsh
    Tcl: .tcl .tk .wish
    Vim: .vim
Reading options from $CTAGS
XXX AT THIS POINT ParsedLeadingOptions=TRUE !!! XXX
Reading initial options from command line

ParsedLeadingOptions is set to TRUE only in the function ParseOptions(),
which is called in quite a lot of places.

So, what's $CTAGS then? According to
http://ctags.sourceforge.net/ctags.html#lbAN:

"CTAGS 
If this environment variable exists, it will be expected to contain a set of
default options which are read when ctags starts, after the configuration
files listed in FILES, below, are read, but before any command line options
are read. Options appearing on the command line will override options
specified in this file. Only options will be read from this file. Note that
all white space in this variable in considered a separator, making it
impossible to pass an option parameter containing an embedded space. If this
is a problem, use a configuration file instead."

What would be the VMS equivalent of a $CTAGS environment variable? In my
case, a symbol:
CTAGS == "$VIM:CTAGS.EXE"

Without that symbol, the following works just fine:
mc DEVICE:[FULL-PATH]CTAGS.EXE --verbose --totals "-L" TMP:sources.txt

Apparently my VMS-symbol CTAGS, that I only (ab)use to specify the location
of ctags.exe, is read as the environment variable $CTAGS, that's supposed to
contain command-line options.

Rrrright. OK, I suppose that's reasonable.

Also, after reading the environment variable $CTAGS, the variable
ParsedLeadingOptions=TRUE, BEFORE (!!!) reading the command line options.
In my opinion, this last fact is a bug, and AFAIK it's not VMS-related.


Hmm... the $CTAGS thing was already implemented in ctags-3.3 (Fri Sep 17
1999). Oh well, I'm sure something has changed since 3.6pre to cause this
problem. Ctags 3.6pre ignores my CTAGS-symbol.


So, how do other people run ctags on VMS then?
From a script?
Using the logical DCL$PATH?
Not at all?
;-)



Coen