[lmi-commits] (no subject)

[Greg Chicares]

branch: master
commit 3c47d2ae1645f575b992d18a52c6088aee32efa6
Author: Gregory W. Chicares <address@hidden>
Date:   Fri Jun 3 16:05:24 2016 +0000

    Don't mark design alternatives as defects [440]
    
    Reading product-file names OAOO is a recognized tradeoff.
    
    Testing for the existence of product files at program startup wouldn't
    be worth the effort, and runtime errors will occur anyway if product
    files exist but are invalid.
---
 ce_product_name.hpp |   10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/ce_product_name.hpp b/ce_product_name.hpp
index f73cd91..d382da8 100644
--- a/ce_product_name.hpp
+++ b/ce_product_name.hpp
@@ -46,18 +46,10 @@
 /// cached, not because of any measured improvement in run time, but
 /// in order to keep the type from mutating with unknown consequences.
 ///
-/// TODO ?? Reconsider this. It seems desirable to recognize changes
+/// SOMEDAY !! Reconsider this. It would be nice to recognize changes
 /// to the data directory made by advanced users, but is that easily
 /// possible without adjusting the base class's 'allowed_' vector and
 /// perhaps changing the way ordinal() works?
-///
-/// TODO ?? Policy-file names are read the first time this class is
-/// instantiated. That's not the best time to do it, because an error
-/// is reported if no such file is found. It would be better to test
-/// that at startup. A simple singleton might be wanted if this
-/// operation should really be performed only once. Alternatively, the
-/// program might write the 'sample' product's files if it finds no
-/// other product.
 
 class ce_product_name
     :public mc_enum_base