Experiment with inverting Regex scan loop

[stephentoub]

The Regex scan loop is currently along the lines of:

while (true)
{
    if (!FindFirstChar()) return false;
    Go();
    if (matched) return true;
}

This is good for cases where matches don't start at the beginning of the string. However, for cases where the match would likely be at the beginning of the input, a form like:

do
{
    Go();
    if (matched) return true;
}
while (FindFirstChar());
return false;

would likely be better, avoiding the extra FindFirstChar call to validate something which Go is already going to validate.

We should evaluate how much this helps one set of cases vs how much this penalizes the other set of cases, and decide whether it's worth switching. (If we do this, note that certain cases in Go currently assume that FindFirstChar has already performed the match, e.g. if the entire expression is a case-sensitive string... that would need to be removed.)

[ghost]

Tagging subscribers to this area: @dotnet/area-system-text-regularexpressions
See info in area-owners.md if you want to be subscribed.

Issue Details

---

The Regex scan loop is currently along the lines of:

while (true)
{
    if (!FindFirstChar()) return false;
    Go();
    if (matched) return true;
}

This is good for cases where matches don't start at the beginning of the string. However, for cases where the match would likely be at the beginning of the input, a form like:

do
{
    Go();
    if (matched) return true;
}
while (FindFirstChar());
return false;

would likely be better, avoiding the extra FindFirstChar call to validate something which Go is already going to validate.

We should evaluate how much this helps one set of cases vs how much this penalizes the other set of cases, and decide whether it's worth switching.

Author:	stephentoub
Assignees:	-
Labels:	area-System.Text.RegularExpressions, tenet-performance, untriaged
Milestone:	7.0.0