|Anonymous | Login | Signup for a new account||2014-08-28 05:12 CEST|
|Main | My View | View Issues | Change Log | Roadmap|
|View Issue Details|
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0005437||OCaml||OCaml general||public||2011-12-21 16:54||2012-09-25 20:07|
|Target Version||Fixed in Version||3.13.0+dev|
|Summary||0005437: Patch to allow command line arguments in Arg module to be undocumented|
The Arg module doesn't have any mechanism for developer command line switches which are not exposed to the user.
This patch hides a switch if its help string begins with the characters "ARG-UNDOC". The documentation string in the .mli is modified to explain this change.
We could, of course, choose something more concise than "ARG-UNDOC" like "*", but then it might clash with existing code.
Coherent Graphics Ltd
|Tags||No tags attached.|
|Attached Files||arg-undocumented.patch [^] (934 bytes) 2011-12-21 16:54 [Show Content]|
First of all, thanks for the patch. However, that looks like quite the hack! Thus, I'm afraid we cannot integrate this patch as-is.
The right way to do this would be to change signature of parse from
val parse :
(key * spec * doc) list -> anon_fun -> usage_msg -> unit
val parse :
(key * spec * doc * documented) list -> anon_fun -> usage_msg -> unit
type documented = bool
However, this would break backwards compatibility. What you can do instead is add an optional unit argument to the parse function and others, that tells whether arguments with no documentation should be printed out.
Thinking about it, that sounds very ad-hoc, and I tend to think you'd be better off keeping a private (tweaked) copy of arg.ml in your source tree...
Maybe the following function could be added:
val parse_rich : (key * spec * rich_doc) -> anon_fun -> usage_msg -> unit
where rich_doc is an abstract type¹ with
val doc : string -> rich_doc
val undocumented : rich_doc
¹: not having "doc" abstract in the first place is what makes it difficult to enrich the behavior now without breaking compatibility.
|Right, but there's a whole lot of functions that take triples (key * spec * doc), are you implying that we should provide an alternative version for each one of them? I can see people asking for more if we were to implement just one...|
After discussing this with others, this has been fixed in svn revision r11939.
To have an option not included in the usage string, just put an empty string for its documentation.
Thanks for this, but the patch isn't right: it conflicts with the padding of help strings. You need to also do:
--- trunk/stdlib/arg.ml 2011-12-22 19:03:46.000000000 +0000
+++ arg.ml 2011-12-23 11:40:17.000000000 +0000
@@ -238,6 +238,7 @@
let add_padding len ksd =
match ksd with
+ | (_, _, "") -> ksd (* Do not pad undocumented ones. *)
| (kwd, (Symbol (l, _) as spec), msg) ->
let cutcol = second_word msg in
let spaces = String.make (len - cutcol + 3) ' ' in
Good catch! I'd never used [align] before so I didn't think of it.
|2011-12-21 16:54||johnwhitington||New Issue|
|2011-12-21 16:54||johnwhitington||File Added: arg-undocumented.patch|
|2011-12-21 17:13||protz||Note Added: 0006467|
|2011-12-21 21:02||gasche||Note Added: 0006472|
|2011-12-22 09:36||protz||Note Added: 0006483|
|2011-12-22 09:36||protz||Assigned To||=> protz|
|2011-12-22 09:36||protz||Status||new => feedback|
|2011-12-22 15:05||protz||Note Added: 0006505|
|2011-12-22 15:05||protz||Status||feedback => resolved|
|2011-12-22 15:05||protz||Fixed in Version||=> 3.13.0+dev|
|2011-12-22 15:05||protz||Resolution||open => fixed|
|2011-12-23 12:44||johnwhitington||Note Added: 0006514|
|2011-12-23 12:44||johnwhitington||Status||resolved => feedback|
|2011-12-23 12:44||johnwhitington||Resolution||fixed => reopened|
|2011-12-23 14:34||protz||Note Added: 0006515|
|2011-12-23 14:34||protz||Status||feedback => resolved|
|2011-12-23 14:34||protz||Resolution||reopened => fixed|
|2012-09-25 20:07||xleroy||Status||resolved => closed|
|Copyright © 2000 - 2011 MantisBT Group|