Re: [BUG] git p4 clone fails when p4 sizes does not return 'fileSize' key

[Luke Diamand <luke@diamand.org>]

On 17 April 2018 at 20:12, Thandesha VK <thanvk@gmail.com> wrote:
> I have few cases where even p4 -G sizes (or p4 sizes) is not returning
> the size value even with latest version of p4 (17.2). In that case, we
> have to regenerate the digest for file save it - It mean something is
> wrong with the file in perforce.

Just to be clear - git-p4 does not use the p4 "sizes" command anywhere AFAIK.

We are just talking about the output from "p4 print" and the
"fileSize" key, right?

Does that happen with the 17.2 version of p4?

> Regarding, sys.stdout.write v/s print, I see script using both of them
> without a common pattern. I can change it to whatever is more
> appropriate.

print() probably makes more sense; can we try to use the function form
so that we don't deliberately make the path to python3 harder (albeit
in a very tiny way).

>
> On Tue, Apr 17, 2018 at 11:47 AM, Mazo, Andrey <amazo@checkvideo.com> wrote:
>> Does a missing "fileSize" actually mean that there is something wrong with the file?
>> Because, for me, `p4 -G print` doesn't print "fileSize" for _any_ file.
>> (which I attribute to our rather ancient (2007.2) Perforce server)
>> I'm not an expert in Perforce, so don't know for sure.

My 2015 version of p4d reports a fileSize.

>>
>> However, `p4 -G sizes` works fine even with our p4 server.
>> Should we then go one step further and use `p4 -G sizes` to obtain the "fileSize" when it's not returned by `p4 -G print`?
>> Or is it an overkill for a simple verbose print out?

If your server isn't reporting "fileSize" then there are a few other
places where I would expect git-p4 to also fail.

If we're going to support this very ancient version of p4d, then
gracefully handling a missing fileSize will be useful.

>>
>> Also, please, find one comment inline below.
>>
>> Thank you,
>> Andrey
>>
>> From: Thandesha VK <thanvk@gmail.com>
>>> Sounds good. How about an enhanced version of fix from both of us.
>>> This will let us know that something is not right with the file but
>>> will not bark
>>>
>>> $ git diff
>>> diff --git a/git-p4.py b/git-p4.py
>>> index 7bb9cadc6..df901976f 100755
>>> --- a/git-p4.py
>>> +++ b/git-p4.py
>>> @@ -2566,7 +2566,12 @@ class P4Sync(Command, P4UserMap):
>>>          relPath = self.stripRepoPath(file['depotFile'], self.branchPrefixes)
>>>          relPath = self.encodeWithUTF8(relPath)
>>>          if verbose:
>>> -            size = int(self.stream_file['fileSize'])
>>> +            if 'fileSize' not in self.stream_file:
>>> +               print "WARN: File size from perforce unknown. Please verify by p4 sizes %s" %(file['depotFile'])
>> For whatever reason, the code below uses sys.stdout.write() instead of print().
>> Should it be used here for consistency as well?
>>
>>> +               size = "-1"
>>> +            else:
>>> +               size = self.stream_file['fileSize']
>>> +            size = int(size)
>>>              sys.stdout.write('\r%s --> %s (%i MB)\n' %
>>> (file['depotFile'], relPath, size/1024/1024))
>>>              sys.stdout.flush()
>>>
>>>
>>> On Tue, Apr 17, 2018 at 10:33 AM, Mazo, Andrey <amazo@checkvideo.com> wrote:
>>>> Sure, I totally agree.
>>>> Sorry, I just wasn't clear enough in my previous email.
>>>> I meant that your patch suppresses "%s --> %s (%i MB)" line in case "fileSize" is not available,
>>>> while my patch suppresses just "(%i MB)" portion if the "fileSize" is not known.
>>>> In other words,
>>>>  * if "fileSize" is known:
>>>>  ** both yours and mine patches don't change existing behavior;
>>>>  * if "fileSize" is not known:
>>>>  ** your patch makes streamOneP4File() not print anything;
>>>>  ** my patch makes streamOneP4File() print "%s --> %s".
>>>>
>>>> Hope, I'm clearer this time.
>>>>
>>>> Thank you,
>>>> Andrey
>>>>
>>>> From: Thandesha VK <thanvk@gmail.com>
>>>>> *I* think keeping the filesize info is better with --verbose option as
>>>>> that gives some clue about the file we are working on. What do you
>>>>> think?
>>>>> Script has similar checks of key existence at other places where it is
>>>>> looking for fileSize.
>>>>>
>>>>> On Tue, Apr 17, 2018 at 9:22 AM, Andrey Mazo <amazo@checkvideo.com> wrote:
>>>>>> Huh, I actually have a slightly different fix for the same issue.
>>>>>> It doesn't suppress the corresponding verbose output completely, but just removes the size information from it.
>>>>>>
>>>>>> Also, I'd mention that the workaround is trivial -- simply omit the "--verbose" option.
>>>>>>
>>>>>> Andrey Mazo (1):
>>>>>>   git-p4: fix `sync --verbose` traceback due to 'fileSize'
>>>>>>
>>>>>>  git-p4.py | 8 ++++++--
>>>>>>  1 file changed, 6 insertions(+), 2 deletions(-)
>>>>>>

Thanks
Luke