Feature database transaction storage #617

rwrzesien · 2018-07-09T12:31:30Z

Resolves #584

Adds GlobalTransactionState and PendingEthereumTransaction models.
Adds DatabaseTransactionsStorage and its usage in ConcentRPC.
Adds get_transaction_count to payments.service.

Marked [WIP] because:

is waiting for answer from related issue about soft delete.
there is some code to uncomment after udpate to SCI 1.4.1 is resolved (link to issue),
it need to be tested on cluster.

Also:
Resolves #605
Resolves #607

kbeker · 2018-07-09T16:04:07Z

concent_api/core/models.py

+    value = CharField(max_length=32)
+
+    to = BinaryField(max_length=20)
+    data = BinaryField()


After testing new SCI I have discovered that data is not BinaryField it is really long string -> probably as a data we will need TextField()

It is string most probably from the same reason as in #617 (comment)

I have decoded the data from sample which you have provided, so I have retrieved the original value from transaction, used it in tests, and it worked fine.

kbeker · 2018-07-09T16:04:52Z

concent_api/core/models.py

+    startgas = CharField(max_length=32)
+    value = CharField(max_length=32)
+
+    to = BinaryField(max_length=20)


Field to in transaction.json is stored as a string.

That's most probably because JSON doesn't support binary data, so it has to be converted to string. I think our point is to be able to store and recreate Transaction to be exactly the same, not storing the data in the same format as JsonTransactionsStorage, but if I am wrong then please correct me. @cameel ?

Btw. 'I think our point is to be able to store and recreate Transaction' is a good scenario for test, I will add it.

You're right. We want to store the original transaction, not a transaction mangled by being converted to JSON.

kbeker · 2018-07-09T16:05:31Z

concent_api/core/models.py

+
+    nonce = DecimalField(max_digits=32, decimal_places=0, unique=True)
+
+    gasprice = CharField(max_length=32)


Gasprice in new SCI is stored as integer

Lets make it Decimal then with max_digits set to 32.

Decimal is fine. But 32 digits is too little. A 32-byte integer has 256 binary digits and the number of decimal digits is about 1/3 of that:

>>> len(str(2 ** 256 - 1)) 78

kbeker · 2018-07-09T16:05:41Z

concent_api/core/models.py

+    nonce = DecimalField(max_digits=32, decimal_places=0, unique=True)
+
+    gasprice = CharField(max_length=32)
+    startgas = CharField(max_length=32)


startgas in new SCI is stored as integer

Lets make it Decimal then with max_digits set to 32.

See my comment above: #617 (comment)

kbeker · 2018-07-09T16:07:03Z

concent_api/core/models.py

+    data = BinaryField()
+
+    v = IntegerField()
+    r = CharField(max_length=32)


Field r in transaction.json is stored as a integer. In my tests i had r field which equaled 105276041803796697890139158600495981346175539693000174052040367753737207356915

I have checked that even BigIntegerField can store numbers in range from -9223372036854775808 to 9223372036854775807 so it is too small for our needs. I think the best would be to convert this number to string and then store as TextField()

Yep, that what I was already pointing out on daily few days ago, BigIntegerField is not enough for this, but Decimal could do in this case, as I now know how sample this data will look exactly in Transaction.

@kbeker I see you have liked TextField very much but believe me it is not a cure for everything ;)

In my tests i had r field which equaled 105276041803796697890139158600495981346175539693000174052040367753737207356915

And that's 78 digits for Decimal field (which requires max_digits parameter) so I would set it safely to 128. Would that be ok @cameel ? Or can we somehow confirm maximum value of it ?

Maximum value for a 32-byte unsigned int is just 2**(32*8)-1. 78 decimal digits is enough.

kbeker · 2018-07-09T16:07:24Z

concent_api/core/models.py

+
+    v = IntegerField()
+    r = CharField(max_length=32)
+    s = CharField(max_length=32)


With s field same story as with r field

Right, I will change it to Decimal too.

kbeker · 2018-07-09T16:09:58Z

concent_api/core/payments/storage.py

+            )
+        except GlobalTransactionState.DoesNotExist:
+            logger.error(f'Trying to get GlobalTransactionState but it does not exist.')
+            raise


I think we will need also method which will get oldest transaction which hasn't been proceed yet. @cameel what do you think?

The nonce does that. By checking its value you can figure out which transaction was added last.

@kbeker Also, we are not implementing our custom interface for this, but follow https://github.com/golemfactory/golem-smart-contracts-interface/blob/v1.4.2/golem_sci/transactionsstorage.py#L16-L53

kbeker · 2018-07-09T16:10:40Z

concent_api/core/payments/storage.py

+                r        = int(ethereum_transaction.r),
+                s        = int(ethereum_transaction.s),
+                to       = encode_hex(ethereum_transaction.to.tobytes()),
+                data     = ethereum_transaction.data.tobytes(),


Well, it's technically against our rules but in this case it clearly improves readability. Maybe we should allow for some exceptions?

@dybi What do you think?

When reviewing this is a lot easier to parse visually

nonce = int(ethereum_transaction.nonce), gasprice = int(ethereum_transaction.gasprice), startgas = int(ethereum_transaction.startgas), value = int(ethereum_transaction.value), v = int(ethereum_transaction.v), r = int(ethereum_transaction.r), s = int(ethereum_transaction.s), to = encode_hex(ethereum_transaction.to.tobytes()), data = ethereum_transaction.data.tobytes(),

than this

nonce = int(ethereum_transaction.nonce), gasprice = int(ethereum_transaction.gasprice), startgas = int(ethereum_transaction.startgas), value = int(ethereum_transaction.value), v = int(ethereum_transaction.v), r = int(ethereum_transaction.r), s = int(ethereum_transaction.s), to = encode_hex(ethereum_transaction.to.tobytes()), data = ethereum_transaction.data.tobytes(),

Yes, exactly, my point here was not to follow old rules but to occasionally break new rules for better readability. But if anyone will insist, I will change it.

@cameel It would be even less readable that in the second part of your example, because there should be no spaces around =.

@rwrzesien. @cameel quite frankly I have really mixed feelings about this . What I mean is - we have pretty strict code style rules and we been struggling for a while to find consensus that all of us would eventually agree upon. I am afraid that having "one exception" here could lead to start the whole discussion again.
My point of view is - if we want to allow some exceptions or loosen up a bit our strict style rules, the best way to do this would be through discussion on retro.

I'd prefer for it to be allowed but you have a point. @rwrzesien please raise this issue on the retro.

kbeker · 2018-07-09T16:10:43Z

concent_api/core/payments/storage.py

+            r           = tx.r,
+            s           = tx.s,
+            data        = tx.data,
+            to          = tx.to,


Same here, done for readability, but if you or anyone insist I will change it.

For me is okay I didn't want to change our style and for me it is more readable :) I have just pointed this because it's not matched with our current style

Please reformat (style)

cameel · 2018-07-09T15:52:42Z

concent_api/core/models.py

+
+    gasprice = CharField(max_length=32)
+    startgas = CharField(max_length=32)
+    value = CharField(max_length=32)


Doesn't PostgreSQL support big integers? If not it would be better to at least use Decimal. We only want integers in this field, not arbitrary strings.

It supports, but big integer is not enough for the data from sample provided by @kbeker . I will switch to Decimal wherever possible.

OK. Decimal is fine.

cameel · 2018-07-09T15:53:23Z

concent_api/core/payments/backends/mock.py

+
+
+def get_transaction_count() -> int:
+    return 10


A null backend should not return arbitrary numbers like this. It should always be zero.

cameel · 2018-07-09T15:55:14Z

concent_api/core/payments/backends/sci_backend.py

@@ -83,6 +83,10 @@ def is_account_status_positive(
    return client_acc_balance > pending_value


+def get_transaction_count() -> int:
+    return ConcentRPC().get_transaction_count()


I did not have a chance to ask earlier: why is it called ConcentRPC? It's strictly related to SCI and Geth and its name should reflect that. RPC = Remote Procedure Call. It's too generic. I would never realize that it's the payment backend/API if I did not know that already.

I actually have no idea, but if we want to change it lets do this in separate issue.

@kbeker Maybe you can explain ?

@cameel Isn't it RemoTe Procedure Call ?

@cameel Isn't it RemoTe Procedure Call ?

Yeah. Fixed my comment.

I actually have no idea, but if we want to change it lets do this in separate issue.

So let's create an issue. Or just a separate pull request. It's a simple thing but it's been bugging me for a long time :)

Created #640

cameel · 2018-07-09T15:55:53Z

concent_api/core/payments/storage.py

+        if not GlobalTransactionState.objects.filter(pk=0).exists():
+            global_transaction_state = GlobalTransactionState(
+                pk=0,
+                nonce=payments_service.get_transaction_count(),  # pylint: disable=no-value-for-parameter


Could you come up with a more general solution for this warning? We don't want to disable it altogether but it should be possible to do this for a specific class.

It is possible for a class, however this is a module. I can add backend = None in all function signatures in core.payments.service, and that will do the job. Is that ok ? backend is added automatically by decorator, however function will crash without backend.

No, we don't want unnecessary defaults if we can avoid it.

What about the status of this in pylint? Are there any chances that they'll fix it soon?

I think not soon according to pylint-dev/pylint#259 (comment)

OK. So let's ignore it.

BTW, I think I have already responded to this. Where did my comment go?

Here :) #617 (comment)

Oh, right :) That's where it went. Sorry :)

cameel · 2018-07-09T15:58:16Z

concent_api/core/payments/storage.py

+            raise Exception(
+                f'Transaction nonce does not match. '
+                f'Current={global_transaction_state.nonce}, '
+                f'Transaction={tx.nonce}'


This should be a custom exception class.

I have followed up the fact that JsonTransactionsStorage is raising pure Exception, and assumed that it might matter, but if you think it doesn't then off course, lets make it a custom exception class.

Let's define a custom exception then. Even if it was checked in any way, a custom exception would inherit from Exception anyway.

We actually should add it to our coding style: Code-Poets/coding-style#10

Well, OK. Let's keep ignoring it for now then.

cameel · 2018-07-09T16:06:47Z

concent_api/core/payments/storage.py

+        try:
+            return GlobalTransactionState.objects.select_for_update().get(
+                pk=0,
+            )


You're never releasing the lock explicitly. From the docs I see that the lock lasts "until the end of the transaction block". The thing is that we will likely have more than one transaction block and they're nested: one explicitly started with @transaction.atomic above and one that's by default on the view (due to ATOMIC_REQUESTS). Could you verify that it's the former that releases the lock? We definitely don't want to keep the row locked until the end of the view.

I have checked it positively however I don't really know how to test it negatively (to cause crash because of lock). Please consider the following code:

with transaction.atomic(using='control'): with transaction.atomic(using='control'): s = Subtask.objects.select_for_update().get(pk=7) s = Subtask.objects.get(pk=7)

It works just fine. Do you think its enough ?

Like this?

with transaction.atomic(using='control'): with transaction.atomic(using='control'): s = Subtask.objects.select_for_update().get(pk=7) s = Subtask.objects.select_for_update().get(pk=7)

This should deadlock if you run it and unless the first atomic() releases the lock.

This code works without errors, so the inner lock is released with the end of inner context manager.

cameel · 2018-07-09T16:09:51Z

concent_api/core/payments/storage.py

+        try:
+            pending_ethereum_transaction = PendingEthereumTransaction.objects.get(
+                nonce=global_transaction_state.nonce - 1
+            )


Your're not decreasing the nonce. This means that you can't for example call revert_last_tx() twice to revert two last transactions.

Is this intentional (maybe it's how SCI works)? Or is it a bug?

My mistake, I thought nonce can never ever be used twice, but I have now checked that JsonTransactionsStorage is doing it.

cameel · 2018-07-09T16:11:25Z

concent_api/core/payments/storage.py

+        """
+        global_transaction_state = self._get_locked_global_transaction_state()
+
+        if global_transaction_state.nonce != tx.nonce:


Shouldn't it be tx.nonce - 1?

Why ? As I understand, the transaction nonce should match 'current' nonce. It is done the same way in https://github.com/golemfactory/golem-smart-contracts-interface/blob/v1.4.2/golem_sci/transactionsstorage.py#L98

But isn't tx the new transaction that already has a nonce set to one higher?

I was assuming that nonce is increased only after the new transaction gets added.

I was assuming that nonce is increased only after the new transaction gets added.

Exactly, so the current transaction should have current nonce, right ? At least it looks like this in JsonTransactionsStorage.

Well, if the JSON version does this like this then it's fine.

cameel · 2018-07-09T16:12:48Z

concent_api/core/tests/test_payments_database_transaction_storage.py

+        pending_ethereum_transaction = PendingEthereumTransaction(
+            nonce=self.global_transaction_state.nonce,
+            gasprice=10 ** 9,
+            startgas=21000,


Please make at least one test that checks the full range of integers. We want to be sure that it does not break with very large numbers.

Ok, but as you see in the discussion in other comments I am still not sure what are maximum values here.

The sizes stated in the issue imply maximum values. For example for a 32-byte unsigned integer the maximum value is 2**(32*8)-1

kbeker · 2018-07-09T16:14:43Z

concent_api/core/models.py

+    v = IntegerField()
+    r = CharField(max_length=32)
+    s = CharField(max_length=32)
+


@cameel @rwrzesien
Don't we need also a field which will tell SCI which transactions was already done and which are awaiting to proceed?

We only need the stuff that's in transactions.json. If there's such a field, please make a comment in the issue and I'll update the description.

No, there is no such field there.

cameel · 2018-07-09T16:16:25Z

Refactors core.payments file structure.

I think it would be better to extract this into a separate pull request so that it does not have to wait until this pull request is merged. I did not see any problems with the refactoring itself so it should go through easily and it's a distraction from the important changes.

rwrzesien · 2018-07-09T23:25:24Z

@cameel

I think it would be better to extract this into a separate pull request so that it does not have to wait until this pull request is merged. I did not see any problems with the refactoring itself so it should go through easily and it's a distraction from the important changes.

Done -> #620

git cherry-pick is a kind of magic.

dybi

Tests seems pretty exhaustive ;) Just two minor style issues.

dybi · 2018-07-12T08:17:42Z

concent_api/core/payments/storage.py

+        """
+        return [
+            Transaction(
+                nonce    = int(ethereum_transaction.nonce),


Please reformat (style)

dybi · 2018-07-12T08:17:47Z

concent_api/core/payments/storage.py

+            r           = tx.r,
+            s           = tx.s,
+            data        = tx.data,
+            to          = tx.to,


Please reformat (style)

rwrzesien · 2018-07-12T11:18:24Z

concent_api/common/singleton.py

-from golem_sci.factory  import new_sci_rpc
+from golem_sci.client import Client
+from golem_sci.contracts.provider import ContractDataProvider
+


Please remove blank line.

rwrzesien · 2018-07-12T11:18:31Z

concent_api/common/singleton.py

+def concent_sci(
+        storage,
+        web3,
+        address,


Fix indentation. If possible add typings. A comment would be nice too.

rwrzesien · 2018-07-12T11:19:54Z

concent_api/common/singleton.py

+        address,
+        tx_sign,
+):
+    if geth_poa_middleware not in web3.middleware_stack:


You can add assert that storage is subclass of TransactionStorage.

rwrzesien · 2018-07-12T11:21:22Z

concent_api/core/payments/storage.py

+        Returns the nonce for the next transaction.
+        """
+        global_transaction_state = self._get_locked_global_transaction_state()
+        return int(global_transaction_state.nonce)


Add method to GlobalTransactionState called get_nonce which will change it to int and reuse it in all places.

rwrzesien · 2018-07-12T11:25:59Z

concent_api/core/exceptions.py

@@ -23,3 +23,7 @@ class GolemMessageValidationError(ConcentBaseException):

 class FrameNumberValidationError(ConcentBaseException):
    pass
+
+
+class TransactionNonceNotMatchError(ConcentBaseException):


I don't think it has to inherit from ConcentBaseException and has error code. It won't be handled by decorator and returned to client. It is internally handled by SCI.

Please also fix the name.

cameel · 2018-07-12T13:35:05Z

concent_api/core/models.py

+
+    v = IntegerField()
+    r = DecimalField(max_digits=128, decimal_places=0)
+    s = DecimalField(max_digits=128, decimal_places=0)


Too many digits.

cameel · 2018-07-12T13:35:21Z

concent_api/core/models.py

+    nonce = DecimalField(max_digits=32, decimal_places=0)
+
+    gasprice = DecimalField(max_digits=32, decimal_places=0)
+    startgas = DecimalField(max_digits=32, decimal_places=0)


Not enough digits.

cameel · 2018-07-12T13:36:00Z

concent_api/core/payments/backends/sci_backend.py

@@ -1,6 +1,8 @@
 from enum import Enum

 from golem_sci.blockshelper import BlocksHelper
+from web3 import Web3
+


If we're doing this, we should add web3 to our dependencies. And remove it later when we stop using it.

Added to requirements.txt

cameel · 2018-07-12T13:37:41Z

concent_api/requirements.lock

@@ -83,4 +90,5 @@ toolz==0.9.0
 typed-ast==1.1.0
 urllib3==1.23
 vine==1.1.4
-web3==3.16.4
+web3==4.2.1
+websockets==5.0.1


Why does SCI depend on websockets?

websockets depends on web3 and is installing during it's installation. Look:
Collecting websockets>=4.0.1 (from web3==4.2.1)

Weird but OK.

cameel · 2018-07-12T13:38:10Z

concent_api/requirements.lock

-eth-keys==0.1.0b4
-eth-tester==0.1.0b15
-eth-utils==0.7.4
+django-constance==2.2.0


Why is the dependency on django-constance being changed here?

As we talked recently on slack and you decided that if something is in pip freeze it should be also in requirements.lock This is how django-constance[database] is installed by pip:

django-constance==2.2.0 django-picklefield==1.0.0

But it's completely unrelated to this pull request. You should make this change separately.

Extracted from this PR

cameel · 2018-07-12T13:39:10Z

concent_api/common/constants.py

@@ -60,6 +60,7 @@ class ErrorCode(enum.Enum):
    QUEUE_WRONG_STATE                                                   = 'queue.wrong_state'
    REQUEST_BODY_NOT_EMPTY                                              = 'request_body.not_empty'
    SUBTASK_DUPLICATE_REQUEST                                           = 'subtask.duplicate_request'
+    TRANSACTION_NONCE_NOT_MATCH                                         = 'transaction.nonce.not_match'


TRANSACTION_NONCE_DOES_NOT_MATCH or TRANSACTION_NONCE_MISMATCH would be a better name.

rwrzesien · 2018-07-13T10:07:16Z

concent_api/common/singleton.py

+    web3: Web3,
+    address: Web3,
+    tx_sign: Callable[[Transaction], None],
+):


You can add -> SCIImplementation

rwrzesien · 2018-07-13T10:09:02Z

concent_api/core/payments/storage.py

+                f'Transaction nonce does not match. '
+                f'Current={global_transaction_state.nonce}, '
+                f'Transaction={tx.nonce}',
+                ErrorCode.TRANSACTION_NONCE_MISMATCH,


It doesn't need error code.

Yes, but won't it be better to have such a thing? :)

But it won't be visible anywhere, it will be just passed to Exception __init__ and disappear.

OK, so I delete it :)

rwrzesien · 2018-07-13T11:49:50Z

concent_api/core/models.py

+    nonce initialized with a value obtained from SCI (Client.get_transaction_count()).
+    """
+
+    nonce = DecimalField(max_digits=32, decimal_places=0, default=0, unique=True)


Shouldn't nonce have max_digits=78 too ?

rwrzesien · 2018-07-13T11:50:25Z

concent_api/core/payments/storage.py

+
+from core.models import GlobalTransactionState
+from core.models import PendingEthereumTransaction
+


Please remove blank line and order alphabetically.

cameel

I have some more remarks but all the really important issues have been addressed already.

cameel · 2018-07-13T14:33:19Z

concent_api/common/singleton.py

+
+#  Temporary SCI factory for Concent needs.
+#  Gets storage as a parameter
+def concent_sci(


Might be a good idea to say in a comment where the original code came from.

cameel · 2018-07-13T14:34:33Z

concent_api/core/models.py

+
+    nonce = DecimalField(max_digits=32, decimal_places=0, default=0, unique=True)
+
+    def save(self, *args, **kwargs):  # pylint: disable=arguments-differ


Why does this warning need to be disabled here?

PILINT gives warning if there is no disable:
Parameters differ from overridden 'save' method (arguments-differ)

So please define it with the same parameters.

cameel · 2018-07-13T14:37:09Z

concent_api/core/models.py

+
+    @property
+    def get_nonce(self):
+        return int(self.nonce)


This sounds like a function name. Properties should be named like variables (i.e. using nouns, not verbs). E.g. nonce_as_int.

Deleted, just as you said in next comment

cameel · 2018-07-13T14:37:36Z

concent_api/core/models.py

+
+    @property
+    def get_nonce(self):
+        return int(self.nonce)


I think that these properties take more code than they actually save. It would be better to just call int() whenever you access a decimal field.

cameel · 2018-07-13T14:39:08Z

concent_api/core/payments/backends/sci_backend.py

-            requestor_address   = requestor_eth_address,
-            provider_address    = provider_eth_address,
+            requestor_address   = Web3.toChecksumAddress(requestor_eth_address),
+            provider_address    = Web3.toChecksumAddress(provider_eth_address),


Wouldn't it be better to let get_forced_payments() do the address conversion internally? It looks like an implementation detail to me.

But, we need to do this conversion before we give addresses as parameters. If not some exception is raised.

But get_force_payments() is our function. If it raises an exception then maybe it shouldn't?

cameel · 2018-07-13T14:41:01Z

concent_api/core/payments/storage.py

+        global_transaction_state = self._get_locked_global_transaction_state()
+
+        if global_transaction_state.get_nonce != tx.nonce:
+            raise TransactionNonceNotMatchError(


Maybe TransactionNonceMismatch would be a better name?

…structor

…seTransactionStorage

rwrzesien requested review from cameel, Jakub89, bartoszbetka, kbeker, pawelkisielewicz, wyspiansky87 and dybi July 9, 2018 12:31

kbeker reviewed Jul 9, 2018

View reviewed changes

cameel suggested changes Jul 9, 2018

View reviewed changes

kbeker reviewed Jul 9, 2018

View reviewed changes

rwrzesien force-pushed the feature-database-transaction-storage branch from d4e5702 to f693332 Compare July 9, 2018 23:05

rwrzesien force-pushed the feature-database-transaction-storage branch 3 times, most recently from b6f6a25 to b23e536 Compare July 10, 2018 01:26

kbeker force-pushed the feature-database-transaction-storage branch 2 times, most recently from 82ce692 to d9afaa3 Compare July 12, 2018 07:32

kbeker changed the title ~~[WIP] Feature database transaction storage~~ Feature database transaction storage Jul 12, 2018

dybi approved these changes Jul 12, 2018

View reviewed changes

rwrzesien commented Jul 12, 2018

View reviewed changes

rwrzesien mentioned this pull request Jul 12, 2018

Rename ConcentRPC to more meaningful name #640

Closed

cameel suggested changes Jul 12, 2018

View reviewed changes

cameel reviewed Jul 12, 2018

View reviewed changes

kbeker force-pushed the feature-database-transaction-storage branch 2 times, most recently from 5fef383 to 2824d6d Compare July 13, 2018 06:48

kbeker force-pushed the master branch from e175336 to b27ae04 Compare July 13, 2018 07:11

kbeker force-pushed the feature-database-transaction-storage branch from 2824d6d to f6890f9 Compare July 13, 2018 07:15

rwrzesien commented Jul 13, 2018

View reviewed changes

kbeker force-pushed the feature-database-transaction-storage branch from d3a7357 to 8d4ce44 Compare July 13, 2018 11:36

rwrzesien commented Jul 13, 2018

View reviewed changes

kbeker force-pushed the feature-database-transaction-storage branch from 8d4ce44 to fade04a Compare July 13, 2018 13:43

cameel approved these changes Jul 13, 2018

View reviewed changes

kbeker force-pushed the feature-database-transaction-storage branch from fade04a to c714b8e Compare July 16, 2018 08:17

[M] Add GlobalTransactionState and PendingEthereumTransaction models

c243ab8

kbeker force-pushed the feature-database-transaction-storage branch from c714b8e to d5f4a97 Compare July 16, 2018 08:44

kbeker and others added 9 commits July 16, 2018 11:06

Update golem_sci requirements to v1.4.1

d094948

Add web3 to requirements.txt

c2bf1d5

Add TransactionNonceMismatch exception

140526c

Add implementation of TransactionsStorage using database

a175117

Add instance of DatabaseTransactionStorage to SCIImplementation's con…

698db70

…structor

Add tests for DatabaseTransactionStorage

17652e6

Add get_transaction_count to payments service and its usage in Databa…

c05bb6d

…seTransactionStorage

Add temporary concent_sci from GSCI

2e38abc

Update sci_backend to GSCI v1.4.1

704d43e

kbeker force-pushed the feature-database-transaction-storage branch from d5f4a97 to 704d43e Compare July 16, 2018 09:07

kbeker merged commit 704d43e into master Jul 16, 2018

kbeker deleted the feature-database-transaction-storage branch July 16, 2018 09:22

cameel added this to the 0.8.0 milestone Jul 31, 2018


		nonce = DecimalField(max_digits=32, decimal_places=0, unique=True)

		gasprice = CharField(max_length=32)


		from core.models import GlobalTransactionState
		from core.models import PendingEthereumTransaction


		nonce = DecimalField(max_digits=32, decimal_places=0, default=0, unique=True)

		def save(self, args, *kwargs): # pylint: disable=arguments-differ

Feature database transaction storage #617

Feature database transaction storage #617

Conversation

rwrzesien commented Jul 9, 2018 • edited by kbeker Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rwrzesien Jul 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rwrzesien Jul 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rwrzesien Jul 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 10, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rwrzesien Jul 10, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 10, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rwrzesien Jul 11, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 10, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cameel Jul 9, 2018 • edited Loading

rwrzesien commented Jul 9, 2018 •

edited by kbeker

Loading

rwrzesien Jul 9, 2018 •

edited

Loading

rwrzesien Jul 9, 2018 •

edited

Loading

cameel Jul 9, 2018 •

edited

Loading

rwrzesien Jul 9, 2018 •

edited

Loading

cameel Jul 9, 2018 •

edited

Loading

cameel Jul 10, 2018 •

edited

Loading

rwrzesien Jul 10, 2018 •

edited

Loading

cameel Jul 10, 2018 •

edited

Loading

rwrzesien Jul 11, 2018 •

edited

Loading

cameel Jul 10, 2018 •

edited

Loading

cameel Jul 9, 2018 •

edited

Loading