[Mono-dev] (no subject)

[Juraj Skripsky]

Hi Vladimir,

Just a small cosmetic detail in "ServerVariablesCollection.cs.patch". In
the switch statement, you could combine quite a few "case"s into one:

switch (name) {
.
.
case "APPL_MD_PATH":
  return _request.WorkerRequest.GetServerVariable ("APPL_MD_PATH");
case "AUTH_PASSWORD":
  return _request.WorkerRequest.GetServerVariable ("AUTH_PASSWORD");
case "CERT_COOKIE":
  return _request.WorkerRequest.GetServerVariable ("CERT_COOKIE");
case "CERT_FLAGS":
  return _request.WorkerRequest.GetServerVariable ("CERT_FLAGS");
.
.

...could be turned into...

switch (name) {
.
.
case "APPL_MD_PATH":
case "AUTH_PASSWORD":
case "CERT_COOKIE":
case "CERT_FLAGS":
  return _request.WorkerRequest.GetServerVariable (name);
.
.


- Juraj


On Thu, 2007-03-29 at 06:04 -0700, Vladimir Krasnov wrote:
> Hello, 
> 
> Please review a performance optimization for HttpRequest.Headers and
> HttpRequest.ServerVariables collections.
> It implements lazy collection fill because most common flow is to ask
> for value by name. Only if user ask for enumerator or count, collection
> will be fully loaded.
> This gives about 25% performance improvement on common flow.
> 
> This causes one test case to fail:
> MonoTests.System.Web.Test_HttpFakeRequest.Test_EmptyUnknownRow :
> NullReferenceException was expected,
> in my opinion this test is incorrect, and should be removed, because it
> tests undocumented behavior that may depend on implementation.
> 
> Vladimir	
> _______________________________________________
> Mono-devel-list mailing list
> Mono-devel-list at lists.ximian.com
> http://lists.ximian.com/mailman/listinfo/mono-devel-list