Reason for duplicate keys on http post? - http

Reading about Http Post on Wikipedia it states that This is a format for encoding key-value pairs with possibly duplicate keys. Is this correct and if so what is the reasoning? Why would a client ever post duplicate keys and if a duplicate key is posted how is the correct corresponding value returned on server side?

To submit multiple values for the same thing.
In PHP, for example, you can name multiple input fields somedata[]. All values of the input boxes are then put in an array named somedata.

Related

Apache NIFI: ExtractAvroMetadata process

The extractAvroMetadata indicates in its properties section that for the propoerty 'Metadata Keys' we can use a comma-seperated list to indicate the fields to get for the avro schema.
Has anyone already used that option since when I mention a list like doc,namesapce it does not work ?
The documentation says this:
"A comma-separated list of keys indicating key/value pairs to extract from the Avro file header. The key 'avro.schema' can be used to extract the full schema in JSON format, and 'avro.codec' can be used to extract the codec name if one exists."
So you can only use keys that are in the header, or the two mentioned keys.

Marketo REST API - what is "dedupeFields" for custom objects?

When it comes to creating/updating custom objects, can I use both dedupeFields or lookupField when pushing the data to Marketo?
What is the difference between the two?
I'm not sure what do you mean under lookupField, as there is no such input field described in the API documentation of the Sync Custom Objects endpoint. (That is the endpoint to create or update custom objects.)
In the other hand, you do not need such a standalone lookup field, as with the input array you provide the objects you want to create or update, with all their important values. Have a look at the sample payload in the docs.
When input is used together with the optional dedupeBy and action fields, you have full control over which object you want to create or update.
Also, the endpoint expects the name of the dedupe field under dedupeBy key, as opposed to dedupeFields. So the name is singular; you can provide a single field name use, and it does what you can expect: if the value in the field for a given record is not unique, an error will be returned for the individual record.

What is the format of the LinkedIn id field r_basicprofile?

We make use of Sign in with LinkedIn for a pre-existing app. The app uses the id field returned as part of the user's profile, however the app has restrictions on what character values can be present in the id.
What are the legal characters that LinkedIn will put in the id?
The description for id says
A unique identifying value for the member.
This value is linked to your specific application. Any attempts to use it with a different application will result in a "404 - Invalid member id" error.
Testing a small sample size, shows things like zHjkl_t-4D, _IcF7_r2b1 and -1ZM8mwCKM, which caused an issue with the field being restricted to starting with alphanumeric characters. I'd like to know the legal values so we can access if LinkedIn signups are suitable for future applications.
Member IDs are presented in Base64 encoded format. Any characters that show up in the Base64 index table are valid.

MVC3 routes - replace id with object name

I'm looking for a fast & elegant way of converting my object IDs with descriptive names, so that my autogenerated routes look like:
/products/oak-table-25x25-3-1
instead of
/products/5bd8c59c-fc37-40c3-bf79-dd30e79b55a5
In this sample:
uid = "5bd8c59c-fc37-40c3-bf79-dd30e79b55a5"
name = "Oak table (25x25) 3/1"
I don't even know how that feature could be named, so that I might google for it.
The problem that I see so far is the uniqueness of that "url-object-name", for example if I have two oak tables 25x35 in the db, and their names differ too little to be uniquely url-named but enough to fool the unique constraint in the db.
I'm thinking of writing that function for name-transform in SQL as an UDF, then adding a calculated field that returns it, then unique-constraining that field.
Is there some more mainstream way of achieving that?
One method is that employed by stackoverflow.com which in your case would be:
/products/5bd8c59c-fc37-40c3-bf79-dd30e79b55a5/oak-table-25x25-3-1
This ensures uniqueness, however the length of the UUID may be a deterrent. You may consider adding a sequential int or bigint identity value to the products table in addition to the uniqueidentifier field. This however would require an additional index on that column for lookup, though a similar index would be required for a Url having only a descritive string. Yet another method would be to use a hash value, seeded by date for instance, which you can compose with the descriptive name. It is simpler to rely on a sequential ID value generated by a database, but if you envision use NoSQL storage mechanisms in the future you may consider using an externally generated hash value to append.
Identity should have 2 properties: it should be unique and unchangable. If you can guarantee, that /products/oak-table-25x25-3-1 will never change to /products/oak-table-25x25-3-1-1 (remember, user can have bookmarks, that shouldn't return 404 statuscode)- you can use name as url parameter and get record by this parameter.
If you can't guarantee uniqueness or want to select record more faster - use next:
/products/123/oak-table-25x25-3-1 - get record by id (123)
/products/123/blablabla - should redirect to first, because blabla no exists or have anoher id
/products/123 - should redirect to first
And try to use more short identities - remember, that at web 2.0 url is a part of UI, and UI should be friendly.
MVC routing (actions) will handle spaces and slashes in a name. It will encode them as %20, and then decode them correctly.
Thus your URL would be /products/oak%20table%2025x25-3%2F1
I have done something very similar in an eCommerce platform I am working on.
The idea is that the URL without the unique ID is better for SEO but we didn't want the unique ID to be the product name that can change often.
The solution was to implement .NET MVC "URL slug only" functionality. The product manager creates "slugs" for every product that are unique and are assigned to products. These link to the product but the product ID and name can be changed whenever.
This allows:
domain.com/oak-table-25x25-3-1
to point to:
/products/5bd8c59c-fc37-40c3-bf79-dd30e79b55a5
(The same functionality can be used on categories too so domain.com/tables can point to domain.com/category/5b38c79c-f837-42c3-bh79-dd405479b15b5)
I have documented how I did this at:
http://makit.net/post/3380143142/dotnet-slug-only-urls

Anyone familiar in Drupal 6 Services module specifically node.save

I can save/update on regular fields but I'm having trouble saving/updating CCK fields. here's an example node.save() XML request - http://pastebin.com/m5ceca16
I'm assuming your XML data mirrors the node object format.
A CCK field 'field_custom' will be accessible via $node->field_custom. Regardless of type and the limit on number of entries, fields are always arrays. If the CCK field only allows one entry, it is $node->field_custom[0].
The indexes below that level depend on the field type. Most, especially numeric and text fields, are 'value' (eg., $node->field_custom[0] = 'foo'). I've used Nodereference fields which use 'nid', from which I would assume Userreference fields use 'uid'.
The structure of your XML seems correct. I would check the structure of a node object on the site (using a var_dump() or the devel module) to make sure all of your array keys and variable names are correct for your field and field type.
2 things to check:
A var_dump of the results of a node_load() doesn't give you the exact format you should use. Your XML must emulate the input format of the node edit form. So while a var_dump might show you several taxonomy terms in an array, the node edit form may expect the taxonomy terms separated by commas. Off-hand I don't see any fields in your example that this would seem to apply to but I mention it anyway.
Your "changed" timestamp must not be in the future, nor must it be too far in the past. The node will not save if this is off by much. This can be an issue if the clocks on one of the computers isn't very accurate. I had an issue where my services server was about 20 seconds behind my services client so all the updates were getting rejected (the server apparently rejected them on the grounds that they were from the future).

Resources