From: John Snow <jsnow@redhat.com>
To: Markus Armbruster <armbru@redhat.com>
Cc: Michael Roth <michael.roth@amd.com>,
Cleber Rosa <crosa@redhat.com>,
qemu-devel@nongnu.org, Eduardo Habkost <ehabkost@redhat.com>
Subject: Re: [PATCH v3 07/16] qapi/expr.py: Add casts in a few select cases
Date: Wed, 24 Feb 2021 17:24:31 -0500 [thread overview]
Message-ID: <9b9b1b1d-1988-2bbd-4f6e-6374ebd3cd88@redhat.com> (raw)
In-Reply-To: <87zgzt1w9w.fsf@dusky.pond.sub.org>
On 2/24/21 7:32 AM, Markus Armbruster wrote:
> John Snow <jsnow@redhat.com> writes:
>
>> Casts are instructions to the type checker only, they aren't "safe" and
>> should probably be avoided in general. In this case, when we perform
>> type checking on a nested structure, the type of each field does not
>> "stick".
>>
>> We don't need to assert that something is a str if we've already checked
>> that it is -- use a cast instead for these cases.
>>
>> Signed-off-by: John Snow <jsnow@redhat.com>
>> Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
>> Reviewed-by: Cleber Rosa <crosa@redhat.com>
>> ---
>> scripts/qapi/expr.py | 8 ++++----
>> 1 file changed, 4 insertions(+), 4 deletions(-)
>>
>> diff --git a/scripts/qapi/expr.py b/scripts/qapi/expr.py
>> index afa6bd07769..f45d6be1f4c 100644
>> --- a/scripts/qapi/expr.py
>> +++ b/scripts/qapi/expr.py
>> @@ -15,7 +15,7 @@
>> # See the COPYING file in the top-level directory.
>>
>> import re
>> -from typing import MutableMapping, Optional
>> +from typing import MutableMapping, Optional, cast
>>
>> from .common import c_name
>> from .error import QAPISemError
>> @@ -232,7 +232,7 @@ def check_enum(expr, info):
>>
>>
>> def check_struct(expr, info):
>> - name = expr['struct']
>> + name = cast(str, expr['struct']) # Asserted in check_exprs
>> members = expr['data']
>>
>
> I believe you need this only so you can declare check_type()'s
> allow_dict to be Optional[str]. More busy-work around allow_dict...
>
> I'm tempted to ask for allow_dict: Any and call it a day.
>
You're right, this is because of the signature for the allow_dict
argument. Ultimately, the pragma is a collection of strings and we need
to prove we are looking up a string somewhere or other.
(Or tell the type checker to leave us alone.)
Unfortunately, the "check" in check_exprs falls off almost immediately.
Working with dictly-typed objects is a little annoying for this reason.
This works for now; but finding a better way to do the pragma checks is
probably the more correct thing. (And not something I really want to try
and get through review until we're done typing, I think.)
>> check_type(members, info, "'data'", allow_dict=name)
>> @@ -240,7 +240,7 @@ def check_struct(expr, info):
>>
>>
>> def check_union(expr, info):
>> - name = expr['union']
>> + name = cast(str, expr['union']) # Asserted in check_exprs
>> base = expr.get('base')
>> discriminator = expr.get('discriminator')
>> members = expr['data']
>
> Likwewise.
>
>> @@ -337,7 +337,7 @@ def check_exprs(exprs):
>> else:
>> raise QAPISemError(info, "expression is missing metatype")
>>
>> - name = expr[meta]
>> + name = cast(str, expr[meta]) # Asserted right below:
>
> "Checked", not "asserted".
>
Oh, yes.
>> check_name_is_str(name, info, "'%s'" % meta)
>> info.set_defn(meta, name)
>> check_defn_name_str(name, info, meta)
>
>
> Cast before check gives me a queasy feeling. It's lying to the type
> checker.
>
> Hamfisted way to avoid lying:
>
> check_name_is_str(expr[meta], ...)
> name = cast(str, expr[meta])
>
> Something like
>
> name = check_name_is_str(name, ...)
>
> might be neater, but then we'd have to find a better function name.
>
OK, I'll look into re-ordering this.
next prev parent reply other threads:[~2021-02-24 22:26 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-23 0:33 [PATCH v3 00/16] qapi: static typing conversion, pt3 John Snow
2021-02-23 0:33 ` [PATCH v3 01/16] qapi/expr.py: Remove 'info' argument from nested check_if_str John Snow
2021-02-23 0:33 ` [PATCH v3 02/16] qapi/expr.py: Check for dict instead of OrderedDict John Snow
2021-02-24 9:30 ` Markus Armbruster
2021-02-24 21:23 ` John Snow
2021-02-25 10:40 ` Markus Armbruster
2021-02-25 20:04 ` John Snow
2021-03-01 16:48 ` Markus Armbruster
2021-02-23 0:33 ` [PATCH v3 03/16] qapi/expr.py: constrain incoming expression types John Snow
2021-02-24 10:01 ` Markus Armbruster
2021-02-24 21:46 ` John Snow
2021-02-25 11:56 ` Markus Armbruster
2021-02-25 20:43 ` John Snow
2021-03-02 5:23 ` Markus Armbruster
2021-02-23 0:33 ` [PATCH v3 04/16] qapi/expr.py: Add assertion for union type 'check_dict' John Snow
2021-02-24 10:35 ` Markus Armbruster
2021-02-24 21:54 ` John Snow
2021-03-24 21:09 ` John Snow
2021-03-25 5:46 ` Markus Armbruster
2021-03-25 19:42 ` John Snow
2021-02-23 0:33 ` [PATCH v3 05/16] qapi/expr.py: move string check upwards in check_type John Snow
2021-02-23 0:33 ` [PATCH v3 06/16] qapi/expr.py: Check type of 'data' member John Snow
2021-02-24 10:39 ` Markus Armbruster
2021-02-24 22:06 ` John Snow
2021-02-25 12:02 ` Markus Armbruster
2021-02-23 0:33 ` [PATCH v3 07/16] qapi/expr.py: Add casts in a few select cases John Snow
2021-02-24 12:32 ` Markus Armbruster
2021-02-24 22:24 ` John Snow [this message]
2021-02-25 12:07 ` Markus Armbruster
2021-02-25 22:10 ` John Snow
2021-02-23 0:34 ` [PATCH v3 08/16] qapi/expr.py: add type hint annotations John Snow
2021-02-24 15:27 ` Markus Armbruster
2021-02-24 22:30 ` John Snow
2021-02-25 12:08 ` Markus Armbruster
2021-02-25 13:56 ` Markus Armbruster
2021-02-25 20:54 ` John Snow
2021-03-02 5:29 ` Markus Armbruster
2021-02-23 0:34 ` [PATCH v3 09/16] qapi/expr.py: Consolidate check_if_str calls in check_if John Snow
2021-02-25 14:23 ` Markus Armbruster
2021-02-25 21:34 ` John Snow
2021-03-02 5:57 ` Markus Armbruster
2021-02-23 0:34 ` [PATCH v3 10/16] qapi/expr.py: Remove single-letter variable John Snow
2021-02-25 14:03 ` Markus Armbruster
2021-02-25 21:56 ` John Snow
2021-02-23 0:34 ` [PATCH v3 11/16] qapi/expr.py: enable pylint checks John Snow
2021-02-23 0:34 ` [PATCH v3 12/16] qapi/expr.py: Add docstrings John Snow
2021-02-23 0:34 ` [PATCH v3 13/16] qapi/expr.py: Modify check_keys to accept any Collection John Snow
2021-02-25 15:41 ` Markus Armbruster
2021-02-23 0:34 ` [PATCH v3 14/16] qapi/expr.py: Use tuples instead of lists for static data John Snow
2021-02-23 0:34 ` [PATCH v3 15/16] qapi/expr.py: move related checks inside check_xxx functions John Snow
2021-02-25 15:28 ` Markus Armbruster
2021-03-25 5:17 ` John Snow
2021-03-25 13:28 ` Markus Armbruster
2021-02-23 0:34 ` [PATCH v3 16/16] qapi/expr.py: Use an expression checker dispatch table John Snow
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=9b9b1b1d-1988-2bbd-4f6e-6374ebd3cd88@redhat.com \
--to=jsnow@redhat.com \
--cc=armbru@redhat.com \
--cc=crosa@redhat.com \
--cc=ehabkost@redhat.com \
--cc=michael.roth@amd.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).