I'm using an app that allows users to upload and download pics. Similar to snapchat they can view the pics of those they follow. After 24 hrs these pics will be moved to an archive table so users will no longer be able to see them. I'm accomplishing this aspect with mysql partitions.
However, on the client side I need to continuously update the mysql query with the date of the last gotten row from the photos table. I store this date on the iOS app. This becomes problematic if the users logs out and allows someone else to log in. I have to clear this data and have not reference point for either user now.
I have a solution to get around this and I want to know how feasible it is. I would create a trigger that would run each time a user retrieved photos. It would update a column on the users table that would hold the last date they viewed. That way when any user logs back in I will have a reference to that last date they viewed. Is this a good idea? I'm open to any suggestions on how to better this approach seeing as how I need to save the pictures instead of just deleting them.
*note the partitions would work but because I need to ensure photos last a minimum of 24 hrs some photos end up lasting more than 24hrs providing the possibility that a users can still see the photos
Photos Table
*id: binary 16
*users_id: foreign: binary 16
*filename: varchar
*created_at: datetime
The photos are stored on Amazon s3
Add a column
visible int not null
Set visible to 1 on upload.
Also put now() into created_at datetime.
Create an event see below:
http://www.mysqltutorial.org/mysql-triggers/working-mysql-scheduled-event/
Have event run daily or every 6 hours, whatever. Can be cron task instead.
When that thing runs it sets visible to 0 for anything over 1 day old.
When users see someone elses profile pictures they see only visible=1 pictures.
The guy that posted the picture sees all his pictures.
So it is automated and you can be asleep at the switch.
Related
We are having a lot of transactions on the site, so for this reason we are re-cycling our booking confirmation numbers/ order id numbers on the confirmation screen which is set into our purchaseID . Since we are re-using our booking confirmation number, in order to make our purchaseID unique we are adding timestamp to our purchaseID variable using pipe delimeter. So formula looks like:
purchaseID = order_id + '|' + timestamp (current date).
My concern here is, let's say I make a booking today and my purchase id looks like -
purchaseID = 5747118 | 6-7-2019
Now I access my confirmation screen again tomorrow and after 2 days, 3 days and so on and I see adobe calls firing. Because I accessed my confirmation page on different dates my timestamp changed and thus my purchaseID is not unique anymore. Even though I am seeing my same booking confirmation page my purchaseID is not unique now. Does this mean, every time i view my confirmation screen on a different day my booking/revenue would be counted multiple times ? If yes, what's the best way to tackle this issue ?
So it sounds like someone can go to your site, make a purchase and see the confirmation page, and then later on, go back to the same confirmation page without actually making another purchase. Maybe they bookmarked the page and come back to it later for reference. Or maybe they refreshed the page, because reasons.
Does your site charge their credit card for accessing the page again? I sure hope not. Your site/coding should be structured in a way that does not keep charging the customer more money every time they view the page again.
And your code logic for outputting Adobe Analytics should be structured in the same way: your coding logic should be that you only output purchase event and variables (e.g. purchaseID) when a purchase actually occurs.
In practice, this sometimes isn't easy to do because of how the site is structured. So part of why purchaseIDexists is to de-duplicate purchases, so that if purchase event and data is re-popped, it will be de-duped. But it only works if you output the same purchaseID when the visitor refreshes the page or otherwise comes back to it later on (where they aren't actually making another purchase).
Which it sounds like you were doing with the original booking confirmation number you pushed to purchaseID. But things went south when you decided to throw a current datestamp into the mix because you started recycling booking confirmation numbers. Well you can't do that. You can use a dynamic value such as the current date/timestamp as part of the value, but you must remember it, and output it in the future.
Maybe this involves adding an extra column to your database with the date/timestamp of purchase (which I have to assume you surely already have), and then pull that value when you pull the booking confirmation number.
Or maybe the solution involves stepping back and rethinking the fact you are recycling booking confirmation numbers. This seems like a bad idea to me. It's definitely a bad idea for your Adobe Analytics implementation, as you have seen for yourself. But is this not a bad idea in general? What happens if a customer buys something today and has # 12345 as proof of transaction to reference, and then tomorrow, a week, a year or whatever from now, some other customer gets the same number?
It stands to reason that you will end up with a mess on your hands, trying to sort out which customer bought what. Transaction ids by their nature are supposed to be forever unique to the transaction. So my very first recommended solution to you would be to stop recycling your booking confirmation numbers. Move to a different format if you need to (e.g. UUID).
Failing that, my next recommendation would be what I said a couple paragraphs up, about storing the date/timestamp in a column at the actual time of purchase (which surely you already have), and then grab and use that value along with the booking confirmation # to use as delimited value, instead of generating the current date on the fly (which absolutely does not work).
I have an iOS social app that uses Firebase as the main database to store all the posts with time stamp included
What I want to achieve is to remove anything that is > 10 days old from my database.
Currently, I am checking this with this super inefficient way (The only way I know). Every time the user queries the firebase, I have swift code that also queries the ENTIRE database and delete all entries that is > 10 days old. This works but it is really inefficient...
What you're trying to do is currently best done on a server you control with a job that runs periodically to scan and delete the old items. You can use the admin SDK for that.
You should also have a index on the time field that you're using to determine how old it is, in order to optimize the query that generates the results.
I have an app I'm working on that is a credits system for a store. A customer brings in items and receives a credit and then can turn around and use that credit towards certain goods in the store. I've set it up so every time a credit holder or credit is created,updated, or destroyed the event is logged. I'm wondering if there is an easy way to use the event data from the logs to create a dashboard displaying things such as X number of credits created and Y number of credits used today. This may not be the right way to go about doing this at all and if so feel free to guide me in another direction. Thanks in advance!
You should save the information into a database (in addition) to the log and operate on it in this fashion.
So for example, maybe you have a User it should be a Model and have credits which should be an integer. You can modify this value every time a transaction happens.
You can also create an associated model 'transactions' which belong_to the user and to find out transactions that happened on a certain day, you would be able to pull up all of the transactions of that user in a certain time range.
If your credits work similar to dollars and money. And your transactions are like orders, you may want to look into using the Spree gem. https://github.com/spree/spree
You definitely do not want to be reading from the logs to do very usual actions like you're describing.
Ok I looked around and couldn't find the answer to this. So the twitter API doesn't make available the date/time a follower became a follower. If I wanted to figure out the number of new followers per day I would some how have to get the previous days and compare the two. Any idea how to do that effectively (like save snap shot every day and save to db or save a snap shot filter everyday save results).
I'd save a snapshot in the DB for comparison. If you store the name of each follower, you could also find out who quits you each day as well as who your new followers are.
I'm working on a Rails application that's kind of like a blog. Users create Entries. I'm trying to work out how to handle time storage and display. I've read this writeup about Rails timezone support.
It's great but it doesn't cover what my app needs to do. It only looks at cases where you want to convert stored time to the current logged in user's time zone. In contrast, the effect I want is...
A user creates an entry in California at 10:00 a.m.
A couple years later he moves to New York and then at some point looks at his old entry. The "created" date should say "10:00 a.m." He doesn't care about time zones. He just wants to know what time of day he felt like it was when he wrote the entry.
If he then edits the Entry in New York the displayed "modified" date is, again, his subjective time of day when he made the edit. (Let's assume he went to "preferences" and changed his time zone setting when he moved.)
Also, for the sake of thoroughness, the app should be able to report the "real" absolute time when an Entry was created or updated.
(Note -- my imaginary user is a guy, but for women it should work roughly the same way.)
The way I'm thinking of implementing it is...
Have the attributes User#time_zone, Entry#created_at_utc, and Entry#updated_at_utc in addition to the standard created_at and updated_at.
The user selects their time zone from a menu when they sign up. (They can change it later if they want.)
The app uses User#time_zone to store created_at and updated_at in the user's subjective local time. If it's 10:00 a.m. for them, the app writes "10:00 a.m." to the DB.
The app also saves the current UTC time in the aforementioned _utc fields to deal with the last requirement above.
Is that a good way to do it? Is there a better way?
The two roads you can take are:
Store a timezone (UTC) in the user account as well as in every post - update the post's timezone along with the updated_at field whenever the user changes the post (if he or she has changed timezones).
Store the timezone only in the user account. When the user changes timezones, update every post that belongs to the user and add/subtract to the created_at/updated_at dates.
The first option seems like the cleanest option to take. For this you would only have to create a new method in your post record:
def locational_updated_at
updated_at + timezone.seconds
end
Where timezone is an integer containing the seconds since UTC.
If you can, you should avoid storing two different sets of timestamps, and you should avoid storing any non-UTC dates. Both of these things will lead to confusion. I'm not completely sure I understand what you're doing (though I like your idea of subjective time), but wouldn't it be enough to just attach a time zone to every post, and always use that zone to display the times? It would default to the time zone set in the author's account, so he could change it when he moved cross-country without affecting previous posts.
I think that's all you need--to attach a time zone to every post. Is that sufficient? Or am I missing some part of this?