# Homework 2021-12-06

### Review: Speed

Consider this code snippet:

useful_client = []

for entry in useful:
    usefulSecurityid = entry['metaData']['metaData']['SECURITY_ID']
    for row in clientSecurities:
        if row['securityId'] == usefulSecurityid:
            break
    if row['securitymaster']['clientspecificsecurity'] is False:
        useful_client.append(entry)

1. Review: Why did we run into problems with this code?
<span style="color:red">1g. The nested loops are an inefficient way to find the matching entry </span>
<span style="color:red">1g.The 'break' pattern only works if there is guranteed a match, else you just end up inserting the final itteration of the loop </span>
1a. Under what circumstances would this code be fine to use?
<span style="color:red">1ag. known finite-length list of client securities with all clientSecurities contains all Useful securities </span>
2. Review: Why would the `semaphore` pattern not (significantly) help this code?
<span style="color:red"> 2g. The number of itterations would be the same </span>
2a. Extra: What is a semaphore?
<span style="color:red"> 2ag. a signaling system to control flow </span>

semaphore = 50

def go(thread):
    if semaphore == 0: 
        sleep(1)
    else:
        old_sem = semaphore
        semaphore = old_sem - 1
        spawn_next_thread()
        do_stuff() # 1/0
        semaphore = semaphore + 1

### Data Access Pipeline
We have a number of data access functions.

$$
\begin{aligned}
get\_url\_transform & : (errorName,date) -> [issue] \\
get\_ultimate\_parent & : clientId \rightarrow clientId \\
get\_account\_data & : [accountId] \rightarrow [accountData] \\
get\_support\_level & : ([securityId], date) \rightarrow [(securityId,supportLevel)] \\
\end{aligned}
$$

1. What is the "type" of `get_client_specific_data`? Extra: Why do each of these have their specific type? ie, why does `get_ultimate_parent` take a single clientId while the others take lists?
<span style="color:red"> 1g. get_client_specific_data([security_ids]) -> [{'securityId','securitymaster'}]
1g.The request/response characteristics of the endpoints i.e. get_ultimate_parent does not return a map of the requested inputs, simply a list of all outputs </span>
2. How does our current data flow work?
<span style="color:red"> 1g. Get all errors, get_support_level for all securities, get account data for all accounts, eliminate client </span>
3. Using the same functions, how can we change our data flow? What would you predict to be the "best" data flow (usually measured by speed)?
4. Code: attempt to code up your data flow. Extra: Prove (or disprove) that you have found a good data flow.

### Look Ahead: Aggregating the data

Begin manipulating the data towards something we can ship to the UI. See the `create_summary_data` in the original script for ideas.

1. What security-level data should we bring to the UI?
2. Code
2a. Attempt to create a list with one entry per security we want to report on.
2b. Begin including data like those in the provided script