Changes between Version 58 and Version 59 of Records


Ignore:
Timestamp:
Feb 26, 2012 7:40:16 PM (2 years ago)
Author:
GregWeber
Comment:

counterpoint: remove redundancy & maintain objectivity

Legend:

Unmodified
Added
Removed
Modified
  • Records

    v58 v59  
    142142=== counterpoint to the above === 
    143143 
    144 I (elaforge) have a project with 314 .hs files, containing 224 `^data .*=` lines, of which 104 define named record fields.  They do tend to collect in large central modules: the largest one has 800 lines and 13 records defined in it.  I have often wished for shorter record names, but I have never wished for two records in the same module with the same field name.  Probably this is a result of the specific application, the example above looks like a database-style app that has lots of IDs instead of directly containing e.g. a Submission or a Conference.  But still, a module per type is overstated, it is one module per type with fields of the same name, and the likelihood of that varies per application. 
     144I (elaforge) have a project with 314 .hs files, containing 224 `^data .*=` lines, of which 104 define named record fields. They do tend to collect in large central modules: the largest one has 800 lines and 13 records defined in it. I have often wished for shorter record names, but I have never wished for two records in the same module with the same field name. Probably this is a result of the specific application, the example above looks like a database-style app that has lots of IDs instead of directly containing e.g. a Submission or a Conference. 
    145145 
    146 I also use qualified imports exclusively, and my conclusion is the opposite of yours: on a large scale they are more important than ever.  Importing is not tedious for me because I use a program to manage imports automatically, which was easy to write only because of qualified imports. 
    147  
    148 Anyway, I believe you that there are kinds of applications that naturally wind up with lots of records in the same module that want to share a name.  But I want to point out that this is not universal, and for it to be compelling evidence rather than anecdote, we'd need a notion of what percentage of programs tend to be this way. 
     146I also use qualified imports exclusively, and my conclusion is the opposite: on a large scale they are more important than ever.  Importing is not tedious for me because I use a program to manage imports automatically, which was easy to write only because of qualified imports.