summaryrefslogtreecommitdiff
path: root/README
diff options
context:
space:
mode:
authorDan Bode <dan@bodepd.com>2010-03-16 00:24:32 -0500
committerDan Bode <dan@bodepd.com>2010-03-16 00:24:32 -0500
commit9c5ff07fc1d45055510798854f4c925e408dda75 (patch)
treed5e2d12b7071be51454e6095d7c8b484c217a9ea /README
parentd3470bcf4629e2bc4830cc82eeed66e8eda11797 (diff)
first version ready for testing
Diffstat (limited to 'README')
-rw-r--r--README22
1 files changed, 22 insertions, 0 deletions
diff --git a/README b/README
index 955a020..c974f03 100644
--- a/README
+++ b/README
@@ -3,3 +3,25 @@ This is a type and a provider for managing sudoers file.
It doesnt work yet, but it will. still trying to learn how parsedfile works and how much I can push its boundaries.
It will update this README when its ready for consumption.
+
+
+KNOWN ISSUES/TODO
+
+ 1. No support for parsing multi-line files (with \). I may add this in the next revision.
+ 2. Forced to use comments to specify NAMEVARS. This means that there is no
+ way to determine if an added line is the same as an existing line.
+ 3. Dependencies are only aplied between records at creation time. It is not possible to insert a new record before an existing one. There is probably a good way to fix this, but I need more time to look into it.
+ 4. A userspec record that starts with Defaults will be parsed as a default. (this may not be possible)
+ 5. I still need to verify combinations of properties that work together and throw exceptions for mismatches.
+ 6. I can probably set up autorequires on aliases in User Spec line, but this order would only be followed if both records were created.
+
+ 7. Performance - seems to be runing to_line way too many times, also flushing way too often
+
+COMMENT NAMEVARS:
+
+I have chosen (or been forced to) implement NAMEVARS using comments. This adds a ton of complexity to the code and its usage. This also means that identical lines will be allowed provided that they have the same namevar.
+
+I have to do this dirty pre-line stuff for user_spec.
+ I had considered requiring uniqueness for the user-spec
+ this does when considering that an alias can also be a user spec user, this violates uniqueness.
+