search

UrbanInstitute / education-data-package-stata

MIT License
19 stars 4 forks source link

speed estimates #23

Closed mchingos closed 6 years ago

mchingos commented 6 years ago

Since the change to number of records per page was made (which seems to really improve performance), the speed estimates seem less accurate than before.

grahamimac commented 6 years ago

Yes, you are right. Because Stata cannot measure time elapsed in a function, I had a hard-coded estimate based on 100 per page. This now needs to be updated based on the 1000 per page limit. I'll update this in the next release. Thanks Matt!

grahamimac commented 6 years ago

Hi Matt, I've "resolved" this in the release coming out tomorrow to have a wider range estimate, because I found some calls take ~1 second per call and others take ~10 seconds. Because Stata cannot measure the time between calls (I've tried, believe me), I can't automatically generate a time estimate for each endpoint. The only alternative to this broad range is to manually hardcode an estimate for each, but this estimate will change every time an update is made, so this is a lot of work. Please re-test on the next release and then we can go from there in terms of resolving this issue.

grahamimac commented 6 years ago

Closing for now, we can reopen if this is not working for people and consider the + - of the manual option.

mchingos commented 6 years ago

Sounds good. Is time roughly a function of the number/type of variables? If so, perhaps the estimate could be coded that way.

From: Graham MacDonald notifications@github.com Sent: Thursday, April 19, 2018 8:34 AM To: UI-Research/education-data-package-stata Cc: Chingos, Matthew; Author Subject: Re: [UI-Research/education-data-package-stata] speed estimates (#23)

Hi Matt, I've "resolved" this in the release coming out tomorrow to have a wider range estimate, because I found some calls take ~1 second per call and others take ~10 seconds. Because Stata cannot measure the time between calls (I've tried, believe me), I can't automatically generate a time estimate for each endpoint. The only alternative to this broad range is to manually hardcode an estimate for each, but this estimate will change every time an update is made, so this is a lot of work. Please re-test on the next release and then we can go from there in terms of resolving this issue.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/UI-Research/education-data-package-stata/issues/23#issuecomment-382717048, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AiqN6xkouWpXmDf06t8GYFLLNxjXA7lxks5tqIRlgaJpZM4TZAKz.