panc: only allow digit-only string as candidate list index (and allow anything else as dict key)#157
Conversation
684704b to
a97075a
Compare
|
We can introduce hex if it is really used, and/or also get rid of octal support (which is confusing imho) |
|
I don't care much about octal/hex notion (I don't remember ever using it), but I have a feeling that relaxing the rules for dict keys would break XML output. The original rules came from the requirements for XML element and attribute names. Automatic escaping would be possible, but that would require changes to the CCM library, and would open the can of worms about how to detect if something is escaped. |
|
Discussion at RAL workshop: OK in principle. @stdweird will additionally drop octal/hex. Then it just needs to be code reviewed & can be merged. Once merged and released then Aquilon can be updated to address: quattor/aquilon#54 |
|
We're removing support for octal, so my vote would be for it to be syntax error rather than silently accepting the value with a different meaning to previous compiler versions. |
|
@stdweird There is no existing problem with aquilon, it only accepts what the PAN compilers accepts. The specific request is for Aquilon to support rooms named |
|
@ned21 ok. but what is your opinion on the octal numbers: syntax error or valid dict key? |
|
Leading zeros are ambiguous (is 09 == 9?) so I agree with @jrha we should make them a syntax error. |
…hing else as dict key)
a97075a to
0da2060
Compare
Fixes #156
This is backwards incompatible because eg hex and/or octal notations cannot be used anymore.