Scratch-vm: Add support for deprecated blocks from 2.0 and 1.4

[TheBrokenRail]

(Edited by @thisandagain on 12/24/2016)

https://wiki.scratch.mit.edu/wiki/List_of_Obsolete_Blocks

[CosmicWebServices]

Umm Arent they trying to get rid of hacked block usage...

[TheBrokenRail]

Yes, but just letting the blocks disappear (like they do currently), I don't believe is a good option, may be we should convert the hacked blocks into normal blocks (like forever if turned into a if block inside a forever block)

[mrjacobbloom]

I like that. And counter could be named counter1 or counter2 etc. using the same method they use to resolve collisions with costume names

[ed-cooper]

This will also need to include the old 1.4 style sound and drum blocks

[CosmicWebServices]

I feel like if they add features that ATers usually have to hack the blocks for that will be key.

[thisandagain]

I'm going to modify the scope the scope of this to be more clear. We need to make sure we support the deprecated opcodes from 1.X to 2.0:
https://wiki.scratch.mit.edu/wiki/List_of_Obsolete_Blocks

For "hacked" blocks, we will support a more robust extension paradigm in 3.0 to be discussed soon.

[towerofnix]

See also #1030.

@thisandagain I assume the particular blocks wanted are those titled "Pseudo-Legacy" in the "Opcode Hacking" section of the slideshow? Are there any others? I've noted a couple extras below anyhow.

---

For reference:

• [x] doForLoop: PR #963.
• [x] doWhile: PR #1032. Its behavior is exactly the opposite of doUntil (i.e. "repeat until"): while the condition is true (rather than false), run the contained blocks.
• [x] warpSpeed ("all at once"): PR #1096. Its behavior is to run the contained blocks as they are (don't actually go into "warp speed", i.e. no-refresh). ("all at once: { move 10 steps, turn 15 degrees }" is functionally equivalent to "move 10 steps, turn 15 degrees".)
• [x] [Counter blocks](https://github.com/LLK/scratch-flash/blob/9fbac92ef3d09ceca0c0782f8a08deaa79e4df69/src/Specs.as#L378-L381): PR #1033.
  
  
  
  • The counter value is a global value that defaults to zero.
  
  
  • counter: Return the counter value.
  
  
  • incr counter: Increment the counter's value by one.
  
  
  • clear counter: Set the counter value to zero.
  
  
• [x] [Scroll blocks](https://github.com/LLK/scratch-flash/blob/9fbac92ef3d09ceca0c0782f8a08deaa79e4df69/src/Specs.as#L386-L391): PR #1099. These are all no-ops.
• [x] ["hide all sprites", "user id"](https://github.com/LLK/scratch-flash/blob/9fbac92ef3d09ceca0c0782f8a08deaa79e4df69/src/Specs.as#L393-L395), and those suspicious "setStretchTo" and "changeStretchBy" blocks in @thisandagain's notes below: PR #1102. Also no-ops.
• [x] ["loud?"](https://github.com/LLK/scratch-flash/blob/96a2a9a7fca2d042da25dc0d4423900163ab4f33/src/Specs.as#L366): PR #1101. Its behavior is to return whether or not the sound level (i.e. result of "loudness") is greater than 10 (so it's identical to loudness > 10).
• Various 1.4 blocks that should automatically be converted to their 2.0+ forms. (Not a checkbox since this will probably be looked at separately from these blocks.)
  
  
  
  • abs (val) -> ("abs") of (val)
  
  
  • sqrt (val) -> ("sqrt") of (val)
  
  
  • stop script -> stop ("this script")
  
  
  • stop all -> stop ("all")
  
  
  • switch to background (bg) -> switch backdrop to (bg)
  
  
  • next background -> next backdrop
  
  
  • forever if (cond): (script) -> forever: if (cond): (script)
  
  

[thisandagain]

@towerofnix That's a great list. Here is a usage breakdown (all shared projects):

We still don't have a 1.4 import / spec-map, but that is something @kchadha will be looking at soon.

[towerofnix]

@thisandagain @kchadha I've made PRs for each block in my checklist, which I've also adjusted a bit since I started it a few weeks ago. I think this is all the relevant blocks; it covers all the ones in thisandagain's breakdown image above, as well as all blocks in the scratch-flash source code.* Are any others wanted?

* Okay, this is just technically a lie :) - I haven't implemented drum:duration:elapsed:from: or midiInstrument:. These are being tracked in #861; they'll take a bit more work to implement (because of the "midi number"-scratch instrument mapping). The rest are here, though!

[thisandagain]

@towerofnix Amazing! Thank you so much. @kchadha and I will take a look at these PRs over the coming week.

[kchadha]

@towerofnix This is awesome! Thanks for making all of these changes. =D

[towerofnix]

Glad to help :)

There's just one more block that needs to be added, which I forgot about -- "costume name" (spotted in this project). I intend for that to be automatically converted from "costume name" to "costume (name)" - i.e. the new costume-(name/number) block - the same way "costume #" is converted to "costume (number)".

I'm just not 100% sure on how to go about doing this. Currently the spec mapping for "costume #" looks like this:

'costumeIndex': {
  opcode: 'looks_costumenumbername',
  argMap: [
  ]
}

There's nothing explicitly saying that the argument should be set to "number", so I'm not sure how to go about making it (in costumeName) be set to "name".

Update - Although if this is too complicated, no worries, there can just be a hidden "costume name" block. (I'm not sure how to approach the issue -- there doesn't seem to be anything in the serialization/sb2.js code that would already help tackle this, and I don't think it's worth it to create an entire new parser feature (that obviously would require tests and what-not) for a single obsolete block.)

[mrjacobbloom]

Along the same lines as "costume name," if the stretch effect from #982 is implemented, should the old "set stretch" blocks to map to those?