Reputation
283
Next privilege 300 Rep.
Create new tags
Badges
3 10
Newest
 Explainer
Impact
~2k people reached

Apr
28
comment Estimate if benchmark sampling difference is significative
In that case, given you don't have the individual samples to filter out the outliers, I'd consider running your tests again and simply drop (or accumulate the sum and sum of squares separately) any results greater than the average + one standard deviation of the data recorded previously.
Apr
28
comment Estimate if benchmark sampling difference is significative
Given most of your data has St Dev > Avg, I'd say you can't learn much from it alone. I'd consider looking at why or how there is so much variation. Specifically, do you really have a wide range of varying results or just some big outliers?
Feb
24
comment Realtime layered time data storage and calculations
Do you have any statuses that are more complex than "if any logins are currently in that state, then that state is current"? If not, I think a "semaphore" concept would work (I don't know your tools, which is why I'm not answering this): atomically increasing a count for the status when the login starts that status and atomically decrementing the count when the login stops being in that status can be used if you can ensure the stopping always gets recorded, even with "unexpected" logouts (e.g. power failure). Obviously, they're in that status whenever the count is positive.
Dec
23
comment Signature-changing decorator: properly documenting additional argument
Without a knowledge of Python, I assume, as it stands, no developer would see the :param comment in any IDE because the association only occurs at runtime. As such I'd suggest a "remarks"-like comment to custom_decorator that mentions there is an extra parameter.
Dec
16
comment Is it reasonable to not write unit tests because they tend to get commented out later or because integration tests are more valuable?
Only adding as a comment because I'm not diligently using unit tests either, but if you were doing "test-driven development" then you should continue with "test-driven maintenance" too :-)
Dec
16
comment What is the exact ingenuity of Unix pipe
And so the "ingenuity of Unix pipe" is the "ingenuity of Unix": all i/o (including interprocess communication, standard files and the rest of the filesystem objects) is handled as files.
Dec
13
comment SDK/DB isDeleted vs Integration Testing
@sleske Well, I did start with that point :-) But the question is still valid for someone who would prefer to not pollute a test database with "deleted" users when running lots of tests under similar conditions.
Dec
9
answered Code attribution to a post quoting an original source
Dec
9
comment Event Sourcing + Updates to multiple aggregates
If the "reserve your seats" option is available, I'd attempt a two-factor version of that: Firstly, before accepting their money, attempt to reserve their booking/seats with a marker that says they haven't paid yet. Then read back these details and confirm they've got the seats/booking without clashes/errors (if not either retry or return to ask to choose details again). Now you know they've got these if the credit check, etc succeeds. If it does, just update the marker to paid. Otherwise, cancel these seats/booking.
Dec
9
comment Workflow, editing things not in your current task
Same conclusion, but it is often the bosses that decide to go for new features and not refactoring :-|
Nov
25
comment How to avoid violating the SRP in a class to manage caching?
I know it's only example code, but use DateTime.UtcNow so you avoid daylight savings changeovers, or even a change in the current timezone.
Oct
27
comment Is it acceptable to define a “module” and “class” within a single .VB element?
@Fabio The big difference its that you still need to reference the name of the static class (for normal methods, not extension methods), where as a VB Module is available as long as it's Namespace is in scope.
Oct
27
revised Is it acceptable to define a “module” and “class” within a single .VB element?
Alternate answer to parenthesized questions.
Oct
27
comment Is it acceptable to define a “module” and “class” within a single .VB element?
@Fabio There are differences, but, for the purposes of this question, yes, C# static class for extension methods could be put in the same file as the main class.
Sep
16
comment Database Structure: A product that will only temporarily be available and then slightly change
To work with your existing structure, I would use the variant concept; it is just that the option for that variant is no longer available to choose.
Sep
16
comment How to detect previous version cookies, and remove them?
Just stating the obvious: the former approach causes a problem if the older version can still be accessed somehow (i.e. the older version will have to drop the newer cookies too); whereas the latter approach will allow the older version to continue to work, with the slight waste of cookie storage on the client with stale cookies hanging around if they don't access the old version again.
Sep
16
answered How do you safely refactor in a language with dynamic scope?
Aug
30
comment Metrics for when to use a hash value?
For reference, the LinqPad test, but it uses my build of DotLisp, which I need to get around to publishing.
Aug
30
comment Metrics for when to use a hash value?
@RobertHarvey I've reproduced that micro-benchmark in LinqPad, and noticed that (1) if you String.Intern the strings first, and so use VB.NET's Is (or C#'s (object)=) you only increase the tip-over point to 5 and (2) if you don't need to otherwise access the list by index, using LinkedList increases the tip-over point to 6 (for normal String =) and about 34 for Interned Strings with Is. What caused me to check the LinkedList was using DotLisp.Cons, with an API designed for interactive interpreter with Object functions had tip-overs at about 5 (=) and about 38 (Is).
Aug
22
answered OPTION STRICT set to off