Revision 322c73366a9198d5bd6be08e91b729c775761821 authored by Diane Gallois-Wong on 31 August 2022, 15:57:02 UTC, committed by Marge Bot on 06 September 2022, 08:21:04 UTC
Notably, remove plugin tests on 1M, since the plugin is no longer responsible for enforcing 1M. Similar tests on 1M already exist in tezt, and will be extended in the next commit to cover all the cases of the removed tests.
1 parent 995112f
tx_rollup_state_repr.ml
(*****************************************************************************)
(* *)
(* Open Source License *)
(* Copyright (c) 2022 Marigold <contact@marigold.dev> *)
(* Copyright (c) 2022 Nomadic Labs <contact@nomadic-labs.com> *)
(* Copyright (c) 2022 Oxhead Alpha <info@oxhead-alpha.com> *)
(* *)
(* Permission is hereby granted, free of charge, to any person obtaining a *)
(* copy of this software and associated documentation files (the "Software"),*)
(* to deal in the Software without restriction, including without limitation *)
(* the rights to use, copy, modify, merge, publish, distribute, sublicense, *)
(* and/or sell copies of the Software, and to permit persons to whom the *)
(* Software is furnished to do so, subject to the following conditions: *)
(* *)
(* The above copyright notice and this permission notice shall be included *)
(* in all copies or substantial portions of the Software. *)
(* *)
(* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR*)
(* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, *)
(* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL *)
(* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER*)
(* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING *)
(* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER *)
(* DEALINGS IN THE SOFTWARE. *)
(* *)
(*****************************************************************************)
open Tx_rollup_errors_repr
type range =
| Interval of {
oldest : Tx_rollup_level_repr.t;
newest : Tx_rollup_level_repr.t;
}
| Empty of {next : Tx_rollup_level_repr.t}
let range_newest = function Interval {newest; _} -> Some newest | _ -> None
let range_oldest = function Interval {oldest; _} -> Some oldest | _ -> None
let extend = function
| Empty {next} -> (Interval {oldest = next; newest = next}, next)
| Interval {oldest; newest} ->
let newest = Tx_rollup_level_repr.succ newest in
(Interval {oldest; newest}, newest)
let shrink = function
| Empty _ -> error (Internal_error "cannot shrink range")
| Interval {oldest; newest} when Tx_rollup_level_repr.(oldest < newest) ->
ok (Interval {oldest = Tx_rollup_level_repr.succ oldest; newest})
| Interval {newest; oldest = _} ->
ok (Empty {next = Tx_rollup_level_repr.succ newest})
let belongs_to range level =
match range with
| Empty _ -> false
| Interval {oldest; newest} ->
Tx_rollup_level_repr.(oldest <= level && level <= newest)
let right_cut range level =
match Tx_rollup_level_repr.pred level with
| None -> ok (Empty {next = Tx_rollup_level_repr.root})
| Some predecessor -> (
match range with
| Interval {oldest; newest = _} when belongs_to range level ->
if Tx_rollup_level_repr.(oldest <= predecessor) then
ok (Interval {oldest; newest = predecessor})
else ok (Empty {next = level})
| _ -> error (Internal_error "cannot cut range"))
let left_extend range level =
match range with
| Interval {oldest = _; newest} -> ok (Interval {oldest = level; newest})
| Empty {next} ->
let newest =
Option.value ~default:level (Tx_rollup_level_repr.pred next)
in
ok (Interval {oldest = level; newest})
let range_count = function
| Empty _ -> 0
| Interval {oldest; newest} ->
Int32.(succ @@ Tx_rollup_level_repr.diff newest oldest |> to_int)
let range_encoding : range Data_encoding.t =
Data_encoding.(
union
[
case
(Tag 0)
~title:"empty"
(obj1 (req "next" Tx_rollup_level_repr.encoding))
(function Empty {next} -> Some next | _ -> None)
(fun next -> Empty {next});
case
(Tag 1)
~title:"interval"
(obj2
(req "newest" Tx_rollup_level_repr.encoding)
(req "oldest" Tx_rollup_level_repr.encoding))
(function
| Interval {newest; oldest} -> Some (newest, oldest) | _ -> None)
(fun (newest, oldest) -> Interval {newest; oldest});
])
let pp_range fmt = function
| Empty {next} -> Format.(fprintf fmt "next: %a" Tx_rollup_level_repr.pp next)
| Interval {oldest; newest} ->
Format.(
fprintf
fmt
"oldest: %a newest: %a"
Tx_rollup_level_repr.pp
oldest
Tx_rollup_level_repr.pp
newest)
type watermark = Tx_rollup_level_repr.t option
let is_above_watermark watermark level =
match watermark with
| Some watermark -> Tx_rollup_level_repr.(watermark < level)
| None -> true
let make_watermark level = Some level
(** The state of a transaction rollup is composed of [burn_per_byte]
and [inbox_ema] fields. [initial_state] introduces their initial
values. Both values are updated by [update_burn_per_byte] as the
rollup progresses.
[burn_per_byte] state the cost of burn per byte to be paid for
each byte submitted to a transaction rollup inbox. [inbox_ema]
is a key factor to impact the update of [burn_per_byte].
[inbox_ema] is the N-block EMA to react to recent N-inbox size
changes. N-block EMA is an exponential moving average (EMA), that
is a type of moving average that places a greater weight and
significance on the most N data points. The purpose of [inbox_ema]
is to get lessened volatility of burn, that is more resistant to
spurious spikes of [burn_per_byte].
The state of the transaction rollup also keeps track of four pointers
to four different rollup levels.
- The [commitment_oldest_level] is the level of the oldest
finalized commitment still stored in the layer-1 storage.
- The [commitment_newest_level] is the level of the most recent
unfinalized commitment in the layer-1 storage.
- The [oldest_inbox_level] is the level of the oldest inbox still stored
in the layer-1 storage.
- The [newest_level] is the level of the most recent inbox in the
layer-1 storage.
*)
type t = {
last_removed_commitment_hashes :
(Tx_rollup_message_result_hash_repr.t * Tx_rollup_commitment_repr.Hash.t)
option;
finalized_commitments : range;
unfinalized_commitments : range;
uncommitted_inboxes : range;
commitment_newest_hash : Tx_rollup_commitment_repr.Hash.t option;
tezos_head_level : Raw_level_repr.t option;
burn_per_byte : Tez_repr.t;
inbox_ema : int;
allocated_storage : Z.t;
occupied_storage : Z.t;
commitments_watermark : watermark;
}
(*
The main use of a transaction rollup state is to keep track of four
pointers to four different rollup levels (see above).
When the rollup is created, these four pointers are initialized with
the [None] value, because no inboxes or commitments have been created
yet. Because inboxes and commitments can be removed from the layer-1
context under certain circumstances, they can be reset to [None].
The state allows us to keep track of three intervals: the finalized
commitments (whose inboxes have been removed from the layer-1
storage), the unfinalized commitments (whose inboxes are still in
the layer-1 storage), and uncommitted inboxes (that is, inboxes
which are still waiting for a commitment).
finalized uncommitted
^^^^^^ ^^^^^^^^
[------------] commitments
[--------------] inboxes
^^^^^^^^
unfinalized
Note that this layout is not the only one that we can witness in
the layer-1 storage, even if it is the more common. It is possible
for instance that there is no unfinalized commitments at a given
time.
finalized
^^^^^^
[----] commitments
[--------------] inboxes
^^^^^^^^^^^^^^^^
uncommitted
Or that we have no more inboxes, but only finalized commitments.
finalized
^^^^^^
CT
[-----] commitments
inboxes
*)
let initial_state ~pre_allocated_storage =
{
last_removed_commitment_hashes = None;
finalized_commitments = Empty {next = Tx_rollup_level_repr.root};
unfinalized_commitments = Empty {next = Tx_rollup_level_repr.root};
uncommitted_inboxes = Empty {next = Tx_rollup_level_repr.root};
commitment_newest_hash = None;
tezos_head_level = None;
burn_per_byte = Tez_repr.zero;
inbox_ema = 0;
allocated_storage = pre_allocated_storage;
occupied_storage = Z.zero;
commitments_watermark = None;
}
let encoding : t Data_encoding.t =
let open Data_encoding in
conv
(fun {
last_removed_commitment_hashes;
finalized_commitments;
unfinalized_commitments;
uncommitted_inboxes;
commitment_newest_hash;
tezos_head_level;
burn_per_byte;
allocated_storage;
occupied_storage;
inbox_ema;
commitments_watermark;
} ->
( ( last_removed_commitment_hashes,
finalized_commitments,
unfinalized_commitments,
uncommitted_inboxes,
commitment_newest_hash,
tezos_head_level,
burn_per_byte,
allocated_storage,
occupied_storage,
inbox_ema ),
commitments_watermark ))
(fun ( ( last_removed_commitment_hashes,
finalized_commitments,
unfinalized_commitments,
uncommitted_inboxes,
commitment_newest_hash,
tezos_head_level,
burn_per_byte,
allocated_storage,
occupied_storage,
inbox_ema ),
commitments_watermark ) ->
{
last_removed_commitment_hashes;
finalized_commitments;
unfinalized_commitments;
uncommitted_inboxes;
commitment_newest_hash;
tezos_head_level;
burn_per_byte;
allocated_storage;
occupied_storage;
inbox_ema;
commitments_watermark;
})
(merge_objs
(obj10
(req
"last_removed_commitment_hashes"
(option
@@ obj2
(req
"last_message_hash"
Tx_rollup_message_result_hash_repr.encoding)
(req
"commitment_hash"
Tx_rollup_commitment_repr.Hash.encoding)))
(req "finalized_commitments" range_encoding)
(req "unfinalized_commitments" range_encoding)
(req "uncommitted_inboxes" range_encoding)
(req
"commitment_newest_hash"
(option Tx_rollup_commitment_repr.Hash.encoding))
(req "tezos_head_level" (option Raw_level_repr.encoding))
(req "burn_per_byte" Tez_repr.encoding)
(req "allocated_storage" n)
(req "occupied_storage" n)
(req "inbox_ema" int31))
(obj1
(req "commitments_watermark" @@ option Tx_rollup_level_repr.encoding)))
let pp fmt
{
last_removed_commitment_hashes;
finalized_commitments;
unfinalized_commitments;
uncommitted_inboxes;
commitment_newest_hash;
tezos_head_level;
burn_per_byte;
allocated_storage;
occupied_storage;
inbox_ema;
commitments_watermark;
} =
Format.(
fprintf
fmt
"cost_per_byte: %a inbox_ema: %d finalized_commitments: %a \
unfinalized_commitments: %a uncommitted_inboxes: %a \
commitment_newest_hash: %a tezos_head_level: %a \
last_removed_commitment_hashes: %a allocated_storage: %a \
occupied_storage: %a commitments_watermark: %a"
Tez_repr.pp
burn_per_byte
inbox_ema
pp_range
finalized_commitments
pp_range
unfinalized_commitments
pp_range
uncommitted_inboxes
(pp_print_option Tx_rollup_commitment_repr.Hash.pp)
commitment_newest_hash
(pp_print_option Raw_level_repr.pp)
tezos_head_level
(pp_print_option (fun fmt (m, c) ->
fprintf
fmt
"(message result: %a, commitment: %a)"
Tx_rollup_message_result_hash_repr.pp
m
Tx_rollup_commitment_repr.Hash.pp
c))
last_removed_commitment_hashes
Z.pp_print
allocated_storage
Z.pp_print
occupied_storage
(pp_print_option Tx_rollup_level_repr.pp)
commitments_watermark)
let adjust_storage_allocation : t -> delta:Z.t -> (t * Z.t) tzresult =
fun state ~delta ->
if Z.(equal zero delta) then ok (state, Z.zero)
else
let occupied_storage' = Z.add state.occupied_storage delta in
if Compare.Z.(occupied_storage' < Z.zero) then
(* returns [Internal_error] if [delta < 0] and [| delta | > state.occupied_storage].
This error should never happen. *)
error
@@ Internal_error
"Storage size should be positive after occupied space is freed."
else
let diff = Z.sub occupied_storage' state.allocated_storage in
if Compare.Z.(diff > Z.zero) then
let state =
{
state with
occupied_storage = occupied_storage';
allocated_storage = occupied_storage';
}
in
ok (state, diff)
else
let state = {state with occupied_storage = occupied_storage'} in
ok (state, Z.zero)
let update_burn_per_byte_helper :
t -> factor:int -> final_size:int -> hard_limit:int -> t =
fun ({burn_per_byte; inbox_ema; _} as state) ~factor ~final_size ~hard_limit ->
let threshold_increase = 90 in
let threshold_decrease = 80 in
let variation_factor = 5L in
let smoothing = 2 in
(* The formula of the multiplier of EMA :
smoothing / (1 + N)
Suppose the period we want to observe is given by the
[factor]. The common choice of smoothing is 2.
*)
let inbox_ema =
inbox_ema + ((final_size - inbox_ema) * smoothing / (1 + factor))
in
let percentage = inbox_ema * 100 / hard_limit in
let computation =
let open Compare.Int in
if threshold_decrease < percentage && percentage <= threshold_increase then
(* constant case *)
ok burn_per_byte
else
Tez_repr.(burn_per_byte *? variation_factor >>? fun x -> x /? 100L)
>>? fun variation ->
let variation =
if Tez_repr.(variation = zero) then Tez_repr.one_mutez else variation
in
(* increase case *)
if threshold_increase < percentage then
Tez_repr.(burn_per_byte +? variation)
else if percentage < threshold_decrease && Tez_repr.(zero < burn_per_byte)
then
(* decrease case, and strictly positive burn *)
Tez_repr.(burn_per_byte -? variation)
else (* decrease case, and burn equals zero *)
ok burn_per_byte
in
match computation with
| Ok burn_per_byte -> {state with burn_per_byte; inbox_ema}
(* In the (very unlikely) event of an overflow, we force the burn to
be the maximum amount. *)
| Error _ -> {state with burn_per_byte = Tez_repr.max_mutez; inbox_ema}
let rec update_burn_per_byte :
t -> elapsed:int -> factor:int -> final_size:int -> hard_limit:int -> t =
fun state ~elapsed ~factor ~final_size ~hard_limit ->
(* factor is expected to be a low number ~ 100 *)
if Compare.Int.(elapsed > factor) then
(* We do not need to compute precisely the new state. *)
{state with burn_per_byte = Tez_repr.zero; inbox_ema = 0}
else if Compare.Int.(elapsed <= 0) then
(* Base case, we take into a account the [final_size] once. *)
update_burn_per_byte_helper state ~factor ~final_size ~hard_limit
else
(* For all the blocks that do not contain inboxes, we act as if
the inbox size was [0]. *)
let state' =
update_burn_per_byte_helper state ~factor ~final_size:0 ~hard_limit
in
let elapsed = elapsed - 1 in
update_burn_per_byte state' ~elapsed ~factor ~final_size ~hard_limit
let has_valid_commitment_at {finalized_commitments; unfinalized_commitments; _}
level =
belongs_to finalized_commitments level
|| belongs_to unfinalized_commitments level
let inboxes_count {unfinalized_commitments; uncommitted_inboxes; _} =
range_count unfinalized_commitments + range_count uncommitted_inboxes
let uncommitted_inboxes_count {uncommitted_inboxes; _} =
range_count uncommitted_inboxes
let commitments_count {finalized_commitments; unfinalized_commitments; _} =
range_count unfinalized_commitments + range_count finalized_commitments
let record_inbox_creation t level =
(match t.tezos_head_level with
| Some tezos_lvl ->
error_when
Raw_level_repr.(level <= tezos_lvl)
(Internal_error "Trying to create an inbox in the past")
| None -> ok ())
>>? fun () ->
let (uncommitted_inboxes, new_level) = extend t.uncommitted_inboxes in
adjust_storage_allocation t ~delta:Tx_rollup_inbox_repr.size
>>? fun (t, diff) ->
ok
( {t with tezos_head_level = Some level; uncommitted_inboxes},
new_level,
diff )
let next_commitment_predecessor state = state.commitment_newest_hash
let finalized_commitment_oldest_level state =
range_oldest state.finalized_commitments
let next_commitment_level state current_level =
match
( range_oldest state.uncommitted_inboxes,
range_newest state.uncommitted_inboxes )
with
| (Some oldest_level, Some newest_level) -> (
if
(* We want to return an error if there is only one inbox in the
storage, and this inbox has been created in the current
block. *)
Tx_rollup_level_repr.(oldest_level < newest_level)
then
(* If [oldest_level < newest_level], we know we are not in
this setup, and we can safely return [oldest_level]. *)
ok oldest_level
else
(* Otherwise, we know that [oldest_level = newest_level], and we
need to check at which Tezos level is has been created. *)
match state.tezos_head_level with
| Some newest_inbox_creation ->
error_when
Raw_level_repr.(current_level <= newest_inbox_creation)
No_uncommitted_inbox
>>? fun () -> ok oldest_level
| None -> error (Internal_error "tezos_head_level was not properly set")
)
| (None, None) -> error No_uncommitted_inbox
| (Some _, None) | (None, Some _) ->
error (Internal_error "rollup state is inconsistent")
let next_commitment_to_finalize state =
range_oldest state.unfinalized_commitments
let next_commitment_to_remove state = range_oldest state.finalized_commitments
let record_inbox_deletion state candidate =
match range_oldest state.unfinalized_commitments with
| Some level when Tx_rollup_level_repr.(candidate = level) ->
shrink state.unfinalized_commitments >>? fun unfinalized_commitments ->
let (finalized_commitments, _) = extend state.finalized_commitments in
ok {state with unfinalized_commitments; finalized_commitments}
| _ -> error (Internal_error "Trying to delete the wrong inbox")
let record_commitment_creation state level hash =
match range_oldest state.uncommitted_inboxes with
| Some oldest ->
error_unless
Tx_rollup_level_repr.(level = oldest)
(Internal_error "Trying to create the wrong commitment")
>>? fun () ->
shrink state.uncommitted_inboxes >>? fun uncommitted_inboxes ->
let (unfinalized_commitments, _) = extend state.unfinalized_commitments in
let state =
{
state with
uncommitted_inboxes;
unfinalized_commitments;
commitment_newest_hash = Some hash;
}
in
if is_above_watermark state.commitments_watermark level then
(* See {{Note inbox}} in [Tx_rollup_commitment_storage] for
why it is safe to “free” the inbox storage when it is
committed too. *)
adjust_storage_allocation state ~delta:(Z.neg Tx_rollup_inbox_repr.size)
>>? fun (state, _) ->
ok {state with commitments_watermark = make_watermark level}
else ok state
| None ->
error (Internal_error "Cannot create a commitment due to lack of inbox")
let record_commitment_rejection state level predecessor_hash =
let unwrap_option msg = function
| Some x -> ok x
| _ -> error (Internal_error msg)
in
let check_none msg = function
| None -> ok ()
| Some _ -> error (Internal_error msg)
in
left_extend state.uncommitted_inboxes level >>? fun uncommitted_inboxes ->
let state = {state with uncommitted_inboxes} in
right_cut state.unfinalized_commitments level
>>? fun unfinalized_commitments ->
match Tx_rollup_level_repr.pred level with
| Some pred_level
when belongs_to state.unfinalized_commitments pred_level
|| belongs_to state.finalized_commitments pred_level ->
(* Case 1. Predecessor level of the rejected commitments has a commitment in the storage *)
unwrap_option "Missing predecessor commitment" predecessor_hash
>>? fun predecessor_hash ->
ok
{
state with
unfinalized_commitments;
commitment_newest_hash = Some predecessor_hash;
}
| Some _ ->
(* Case 2. Predecessor level of the rejected commitments has its
commitment removed from the storage *)
check_none "Unexpected predecessor hash" predecessor_hash >>? fun () ->
unwrap_option
"Missing commitment hash"
state.last_removed_commitment_hashes
>>? fun (_, pred_hash) ->
ok
{
state with
unfinalized_commitments;
commitment_newest_hash = Some pred_hash;
}
| None ->
(* Case 3. The rejected commitment is the commitment of the root level *)
ok {state with unfinalized_commitments; commitment_newest_hash = None}
let record_commitment_deletion state level hash message_hash =
match range_oldest state.finalized_commitments with
| Some oldest when Tx_rollup_level_repr.(level = oldest) ->
shrink state.finalized_commitments >>? fun finalized_commitments ->
ok
{
state with
finalized_commitments;
last_removed_commitment_hashes = Some (message_hash, hash);
}
| _ -> error (Internal_error "Trying to remove an incorrect commitment")
let burn_cost ~limit state size =
Tez_repr.(state.burn_per_byte *? Int64.of_int size) >>? fun burn ->
match limit with
| Some limit when Tez_repr.(limit >= burn) ->
error (Submit_batch_burn_exceeded {burn; limit})
| _ -> ok burn
let finalized_commitments_range state =
match
( range_oldest state.finalized_commitments,
range_newest state.finalized_commitments )
with
| (Some oldest, Some newest) -> Some (oldest, newest)
| _ -> None
let check_level_can_be_rejected state level =
match
( range_oldest state.unfinalized_commitments,
range_newest state.unfinalized_commitments )
with
| (Some oldest, Some newest) ->
error_unless Tx_rollup_level_repr.(oldest <= level && level <= newest)
@@ Cannot_reject_level
{provided = level; accepted_range = Some (oldest, newest)}
| _ -> error @@ Cannot_reject_level {provided = level; accepted_range = None}
let last_removed_commitment_hashes state = state.last_removed_commitment_hashes
let head_levels state =
match (state.uncommitted_inboxes, state.tezos_head_level) with
| (Empty {next = l}, Some tz_level) ->
Option.map (fun l -> (l, tz_level)) (Tx_rollup_level_repr.pred l)
| (Interval {newest; _}, Some tz_level) -> Some (newest, tz_level)
| _ -> None
module Internal_for_tests = struct
let make :
?burn_per_byte:Tez_repr.t ->
?inbox_ema:int ->
?last_removed_commitment_hashes:
Tx_rollup_message_result_hash_repr.t * Tx_rollup_commitment_repr.Hash.t ->
?finalized_commitments:Tx_rollup_level_repr.t * Tx_rollup_level_repr.t ->
?unfinalized_commitments:Tx_rollup_level_repr.t * Tx_rollup_level_repr.t ->
?uncommitted_inboxes:Tx_rollup_level_repr.t * Tx_rollup_level_repr.t ->
?commitment_newest_hash:Tx_rollup_commitment_repr.Hash.t ->
?tezos_head_level:Raw_level_repr.t ->
?occupied_storage:Z.t ->
?commitments_watermark:Tx_rollup_level_repr.t ->
allocated_storage:Z.t ->
unit ->
t =
fun ?(burn_per_byte = Tez_repr.zero)
?(inbox_ema = 0)
?last_removed_commitment_hashes
?finalized_commitments
?unfinalized_commitments
?uncommitted_inboxes
?commitment_newest_hash
?tezos_head_level
?(occupied_storage = Z.zero)
?commitments_watermark
~allocated_storage
() ->
let to_range = function
| Some (oldest, newest) ->
assert (Tx_rollup_level_repr.(oldest <= newest)) ;
Interval {oldest; newest}
| _ -> Empty {next = Tx_rollup_level_repr.root}
in
let unfinalized_commitments = to_range unfinalized_commitments in
let finalized_commitments = to_range finalized_commitments in
let uncommitted_inboxes = to_range uncommitted_inboxes in
{
last_removed_commitment_hashes;
burn_per_byte;
occupied_storage;
allocated_storage;
inbox_ema;
finalized_commitments;
unfinalized_commitments;
uncommitted_inboxes;
commitment_newest_hash;
tezos_head_level;
commitments_watermark;
}
let get_inbox_ema : t -> int = fun {inbox_ema; _} -> inbox_ema
let get_occupied_storage : t -> Z.t =
fun {occupied_storage; _} -> occupied_storage
let set_occupied_storage : Z.t -> t -> t =
fun occupied_storage st -> {st with occupied_storage}
let get_allocated_storage : t -> Z.t =
fun {allocated_storage; _} -> allocated_storage
let set_allocated_storage : Z.t -> t -> t =
fun allocated_storage st -> {st with allocated_storage}
let reset_commitments_watermark : t -> t =
fun st -> {st with commitments_watermark = None}
let get_commitments_watermark : t -> Tx_rollup_level_repr.t option =
fun st -> st.commitments_watermark
end
Computing file changes ...