Author Mark.Shannon
Recipients Mark.Shannon, pablogsal, rhettinger, serhiy.storchaka
Date 2020-12-22.12:02:51
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1608638571.52.0.00224943237142.issue42718@roundup.psfhosted.org>
In-reply-to
Content
While the impact of making `if 0` and `while True` appear when tracing can be mitigated, the impact of `continue` is more of a concern.

The following loop:

while True:
    if test:
        continue
    rest

PEP 626 requires that the `continue` is traced, and continue can occur once per iteration. So inserting a NOP for a continue will have a measurable impact on performance.

In some cases the NOP can be folded into the preceding or following bytecode, but often it cannot because the code is both branchy and spread across several lines.

If PEP 626 allowed zero-width entries in the line number table, then any remaining NOPs could be eliminated in the assembler, at the cost of a little additional complexity in `maybe_call_line_trace()`
History
Date User Action Args
2020-12-22 12:02:51Mark.Shannonsetrecipients: + Mark.Shannon, rhettinger, serhiy.storchaka, pablogsal
2020-12-22 12:02:51Mark.Shannonsetmessageid: <1608638571.52.0.00224943237142.issue42718@roundup.psfhosted.org>
2020-12-22 12:02:51Mark.Shannonlinkissue42718 messages
2020-12-22 12:02:51Mark.Shannoncreate