Include bytes sent/received in HTTP requests for replays #7373

Activity

1. bruno-garcia

   added

   Type: Improvement

   Package: replay

   on Mar 7, 2023
2. bruno-garcia commented on Mar 7, 2023

   

   bruno-garcia

   on Mar 7, 2023

   MemberAuthor

   More actions

   FYI @krystofwoldrich outside of replays this would affect React Native and bring in on par to the other mobile SDKs if we're aligning.
3. mydea commented on Mar 8, 2023

   

   mydea

   on Mar 8, 2023

   Member

   More actions

   So just to clarify, we'll add this for all breadcrumbs, not just in replay itself. We'll add:

   • request_body_size
   • response_body_size

   Which we get the same way as for replay recording length:

   new TextEncoder().encode(recordingData).length

   (Which is not supported in IE11, so we'll just not have this data there).

   Final question: Should we do this for all requests/responses, or just for e.g. JSON? For responses we can probably look at content-type headers, for requests not so much (we can't access them).

   I would propose one of these options, for simplicity:

   1. Only capture it if we detect it to be a JSON string (starts with { and ends with }, or starts with [ and ends with ]), or form data (TBD how to detect this)
   2. Always capture it from the text representation (not 100% sure but may be incorrect for non-text media types, I guess)
4. krystofwoldrich commented on Mar 8, 2023

   

   krystofwoldrich

   on Mar 8, 2023

   Contributor

   More actions

   @bruno-garcia Thanks
   @mydea RN doesn't have TextEncoder either. If possible let's use the TextEncoderInternal abstraction so we can supply our own.
5. billyvg commented on Mar 8, 2023

   

   billyvg

   on Mar 8, 2023

   Member

   More actions

   Final question: Should we do this for all requests/responses, or just for e.g. JSON? For responses we can probably look at content-type headers, for requests not so much (we can't access them).

   Shouldn't we be able to access the request headers (at least for fetch?)

   I would propose one of these options, for simplicity:

   1. Only capture it if we detect it to be a JSON string (starts with { and ends with }, or starts with [ and ends with ]), or form data (TBD how to detect this)
   2. Always capture it from the text representation (not 100% sure but may be incorrect for non-text media types, I guess)

   I'd lean towards #2, but we should do some testing to see how inaccurate this can be.
6. billyvg

   changed the title [-]Include bytes sent/received in HTTP requests[/-] [+]Include bytes sent/received in HTTP request breadcrumbs[/+] on Mar 8, 2023
7. billyvg

   changed the title [-]Include bytes sent/received in HTTP request breadcrumbs[/-] [+]Include bytes sent/received in HTTP requests for replays[/+] on Mar 9, 2023
8. billyvg

   assigned

   mydea

   on Mar 14, 2023
9. mydea

   closed this as completedin #7407on Mar 16, 2023
10. dprevost-LMI commented on Nov 28, 2024

    

    dprevost-LMI

    on Nov 28, 2024

    Last edited by dprevost-LMI

    More actions

    bytes or kb, bytes I assume?

    👍React with 👍1Reacted by Jake Lynch