conlini / h2h Goto Github PK

Need utility scripts for common functions such as DB upgrades, static files publish, etc

This would reduce the need for dev/deployment to not have to login in to a server and ease setup

to begin with, we need scripts/utility for

1. DB upgrade
2. static file publish
3. template generation

List of quick wins/wish list

1. Cache product items on start in repo. Use the same for filtering
2. Create "bucket filters" for recently used filters
3. Have a bootup/static data ingest script
4. Compile/Publish templates script
5. Cache refresh at configurable time
6. Logging configuration
7. Add metrics of time taken for requests. log time taken, request type to begin with

most of the code is filled with either using "param_name" and the rest uses "property_name"

This causes immense confusion. Clean up the same and keep single usage terminology

Caching of product items doesnt help caching with a subsequent call to its foreign referenced object

a call such item.itemparam_set.all() would land up hitting the DB once again for each ItemParam and hence defeats the overall caching process

Proposal: Convert the DB models to an internal app model when loading and vice versa when saving

Abstract
Build/Suggest me a stack works on the principle of asking user basic questions on what they wish to do and what are some of basic tech requirements. The h2h engine can then use data from various tech products to suggest a suitable stack for the user.
The Build me a stack can further go on the deploy a bare minimum setup based on suggested technologies on the users account and provide them with end points to interact with

Proposed Framework
I am tinkering with the idea of using workflows for this feature. Based on an entry point(say a static 'no clue what you want' node') the user is given questions. Each question is embedded with hints on what the target tech and filters would correspond to, also keeping in mind previous answers. This would then lead the user down one of the paths of the workflow and ultimately providing the required stack

Example

Do you wish to use NoSql | Yes/No
If yes:
    [options] : Couch/Mongo/Cassandra/etc?
    [Next - question ] Would you require attribute level filtering capabilities | Yes/No
        if yes:
           [options] couch/mongo
        else:
           [options] cassandra
else:
    [options] mysql, oracle, postgres

The evaluator would pick hints to provide the next possible path.
It also will allow easily adding new flows into the system as the product matures

@rahulmamgain @heckdevice Let me know if this makes sense and I can then try to build a framework with some norms on the same

Abstract
With a recent patch I have now moved to hierarchical templates within the h2h.
These templates are rather static in the context of "true" templates and only use the concept of inheritance and blocks

Motivation
The notion of having to define the entire page structure for a new page was quite annoying. Templates allows defining the base structure and then subsequent user views(home, category entry, product entry, etc) to inherit the template and substitute the basic view functionality within the templates body block.
This allowed for a single base.html to house all the styles and js required for the sites functionality and individual pages to concentrate/leverage on the same. Thus a change in the base look nd feel would require a localized change

Working Agreement
The usage of templates would be purely static and dev phase only.
The ultimate idea would be to tie in generator scripts that would deploy each of the rendered html into a set of generated folders as static resource
Note: the templates would not be generated on run-time/page request, but as part of a deployment/build phase

Issues/Concerns
The concept of having python based templates means we now will land up in probably slower dev of the UI.
UI dev would require a cycle of [ deploy generated html > test > modify template ]
In my view this could be a big hurdle and scope of slowing down the quick feedback that non-template/fully qualified HTML pages will have

@rahulmamgain Does this make sense in pursuing or is there an equivalent way of doing this with HTML like templates(without moving to JSPs) and not compromise on dev ease?

Let me know your thoughts

Adding a value to the range filter appends the filter value as a string rather than an int

{"query":{"category":"cat1","filters":[{"active dev":[true]},{"read speed":["900","1000"]},{"released":[854,"1000"]}]}}

The API expects these to be an int

This is on branch feature/task/tests