1033394 - Switch bzexport to using the native Bugzilla REST API

Status: RESOLVED FIXED

(Developer Services :: Mercurial: bzexport, defect)

Description

[Ed Morley [:emorley]]

bzexport currently uses:
https://api-dev.bugzilla.mozilla.org/latest/

The new endpoint, which "should" be identical is:
https://bugzilla.mozilla.org/bzapi/

Ted already landed the switch (as a no bug):
https://hg.mozilla.org/hgcustom/version-control-tools/rev/782f314c73d4

However this caused bug 1033258, so I had to revert in:
https://hg.mozilla.org/hgcustom/version-control-tools/rev/30b6d858ed1c

We should figure out what needs fixing in the compatibility layer so that both bzexport and other projects can switch to it without other changes (if possible).

Comment 1

[Gregory Szorc [:gps]]

PSA: There is a new Python client API for the REST API: https://github.com/AutomatedTester/Bugsy. The package is checked into the version-control-tools repo. It still has a ways to go before it can support bzexport's needs. But I'd like to think bzexport's custom Bugzilla Python module is going the way of the dodo.

Comment 2

[(not currently active) Ted Mielczarek]

Sorry, guess I should have filed a bug to track that! Looks like we can re-land this.

Comment 3

[(not currently active) Ted Mielczarek]

glob says the production push that will pick up bug 1033258 should be in ~12 hours, so if nobody beats me to it I'll re-land tomorrow.

Comment 4

[:glob ✱]

(In reply to Ted Mielczarek [:ted.mielczarek] from comment #3)
> glob says the production push that will pick up bug 1033258 should be in ~12
> hours, so if nobody beats me to it I'll re-land tomorrow.

this is now live.

Comment 5

[(not currently active) Ted Mielczarek]

http://hg.mozilla.org/hgcustom/version-control-tools/rev/1c20d2343aee

Comment 6

[Ed Morley [:emorley]]

Reverted since review request comments are missing from bugmails when using the native APIs, whereas they are not when using the legacy bzapi (see bug 1054181 & bug 508541). This issue is also affecting reviewboard integration.

https://hg.mozilla.org/hgcustom/version-control-tools/rev/dfb611fc22be

Comment 7

[Mike Hommey [:glandium]]

This was also causing the following when attaching new version of patches:
Error: There is no flag with the id '968857'.
abort: Could not update attachment 8483868 [details] [diff] [review]: HTTP Error 404:

Comment 8

[Mark Côté [:mcote]]

The blocker bugs here should be fixed now.  Can we try this again?

Comment 9

[Gregory Szorc [:gps]]

The version-control-tools repo now supports running BMO (via Docker) as part of tests. We can actually write reasonable tests for bzexport now. I would encourage this since bzexport's Mercurial extension currently has 0% test coverage.

Comment 10

[Gregory Szorc [:gps]]

Attachment: template for mercurial tests

Here is a template for a bzexport test using BMO running in Docker. The
test fails, possibly because the Dockerized BMO isn't exposing bzapi?
I'm not sure and I have other priorities to look at now (reviewboard).

Comment 11

[Gregory Szorc [:gps]]

I really wish bzexport had a field for "don't assign bug when uploading"

Comment 12

[Ed Morley [:emorley]]

(In reply to Gregory Szorc [:gps] from comment #11)
> I really wish bzexport had a field for "don't assign bug when uploading"

--no-take-bug iirc

Comment 13

[Ed Morley [:emorley]]

Attachment: Switch to the new bzapi compatibility layer endpoint

Comment 14

[Ed Morley [:emorley]]

Attachment: switch to the new bzapi compatibility layer endpoint

Comment 15

[Ed Morley [:emorley]]

Attachment: switch to the new bzapi compatibility layer endpoint

Comment 16

[Ed Morley [:emorley]]

Attachment: more debugging

Comment 17

[Ed Morley [:emorley]]

Comment on attachment 8529022 [details] [diff] [review]
more debugging

(Sorry for the noise, trying to get a set of STR for bug 1105122)

Comment 18

[Ed Morley [:emorley]]

(In reply to Mike Hommey [:glandium] from comment #7)
> This was also causing the following when attaching new version of patches:
> Error: There is no flag with the id '968857'.
> abort: Could not update attachment 8483868 [details] [diff] [review]: HTTP
> Error 404:

I've tracked this down to bug 1105433.
We'll need to wait until that's fixed before trying to land this again.

Comment 19

[Gregory Szorc [:gps]]

https://hg.mozilla.org/hgcustom/version-control-tools/rev/434e5c9f14b2

Comment 20

[Gregory Szorc [:gps]]

I just added some .t tests for bzexport to version-control-tools. They run Bugzilla inside Docker. If we write enough tests, we'll uncover bugs in bzapi. I may very well go on a bzexport test-writing binge later today.

Comment 21

[Gregory Szorc [:gps]]

I'm about to make some people happy.

Comment 22

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bz://1033394/gps

Comment 23

[Gregory Szorc [:gps]]

/r/2129 - tests: dump more bug state
/r/2131 - bzexport: test that uploading a replacement patch works
/r/2133 - bzexport: add test for reviewer selection
/r/2135 - bzexport: partially implement reviewer switching test
/r/2137 - bzexport: enable loading modules from version-control-tools
/r/2139 - bzexport: use win_get_folder_path from mozhg.auth
/r/2141 - bzexport: use profile path finding from mozhg.auth
/r/2143 - bzexport: use profiles loading code from mozhg.auth
/r/2145 - bzexport: use mozhg.auth for finding cookies from profile
/r/2147 - bzexport: use mozhg.auth.getbugzillaauth
/r/2149 - bzexport: support for using a requests.Session for making requests
/r/2151 - bzexport: use REST API to obsolete attachments (bug 1033394)

Pull down these commits:

hg pull review -r f74d43e4dd47d040b80f8b2015c78eb515921e33

Comment 24

[Gregory Szorc [:gps]]

https://reviewboard.mozilla.org/r/2127/#review1371

A few comments on the state of the series.

First, I just noticed some local test failures due to expected bug status mismatch. Lots of `CONFIRMED` should be `UNCONFIRMED`.

Second, I should probably write more tests around bzexport auth.

Third, I should probably deprecate the `bzexport.username` config variables. I imagine some people have username but no password defined. When we switch to `mozhg.auth`, we'll look in `bugzilla.username` instead of `bzexport.username`. Upon finding nothing there, we'll prompt for it. Users will be like "wtf, I have the username defined." I like pro-actively notifying users of deprecation. Alternatively, we could teach `mozhg.auth` about pre-defined username values. But that would mean supporting `bzexport.username` indefinitely. I'd prefer we encourage people to switch to `bugzilla.username`, as that is used by the `reviewboard` extension as well and saves people from having multiple definitions.

FWIW, the `mozhg.auth` code is already extensively tested. We just need to review the interaction with bzexport.

Comment 25

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2135/#review1373

++ for this :-)

Comment 26

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2147/#review1377

::: hgext/bzexport/bzauth.py
(Diff revision 1)
> -    if not username:
> -        username = ui.prompt("Enter username for %s:" % bugzilla, default='')
> -    if not password:
> -        password = ui.getpass("Enter password for %s: " % username)

The equivalent logic in getbugzillaauth() only prompts for the password if the username isn't set (since the conditionals are nested over there). ie: it's no longer possible to set just the username in the hgrc, and enter the password by prompt. 

If this is intentional, that's fine - just wanted to call it out in case not :-)

Comment 27

[Ed Morley [:emorley]]

Comment on attachment 8545777 [details]
MozReview Request: bz://1033394/gps

Thank you for doing this :-)

Comment 29

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2147/#review1389

> The equivalent logic in getbugzillaauth() only prompts for the password if the username isn't set (since the conditionals are nested over there). ie: it's no longer possible to set just the username in the hgrc, and enter the password by prompt. 
> 
> If this is intentional, that's fine - just wanted to call it out in case not :-)

That seems bad -- IMHO, saving passwords in config files, especially if they're unencrypted, is awful and should be heavily discouraged (yet still allowed). Having just the username in the config file is totally fine, on the other hand (even though it's crap from a usability POV to enter the password constantly.) So disallowing just-username is an encouragement to put a password in the config file.

Comment 30

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2149/#review1391

Wow this UI is confusing.

Comment 31

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2145/#review1393

Er... this patch seems to remove the old code and import the new code with a new name, but I didn't see where it switched to it. Oh well, I will just assume this patch series converges.

Comment 32

[Gregory Szorc [:gps]]

https://reviewboard.mozilla.org/r/2145/#review1395

mozhg.auth has existed for a while. It was written as part of the MozReview work. The code is heavily based on bzexport.

Comment 33

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2143/#review1397

Hrm. This returns the first profile. Unless you're sorting them in a useful way, this is going to break my setup. Though the comments seem to indicate that this may be a temporary state, and eventually you'll search all profiles. I'll keep reading.

Comment 34

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2141/#review1399

It's a little weird to give r+ based on switching to mysterious code I haven't read, but I'm lazy and it's tested, so whatever.

Comment 35

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2139/#review1401

It's sad that the new code is identical, since it's ugly code. Oh well.

Comment 36

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2137/#review1403

Now running a mystery file ../bootstrap.py, after importing things. La de da.

Comment 37

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2135/#review1405

There's no way to mark a test as an expected failure?

(And am I now re-reviewing stuff edmorley has r+'d? RB is so confusing...)

Comment 38

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2133/#review1407

Comment 39

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2131/#review1409

Comment 40

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2129/#review1411

It might be nice to have options for what to include, but this doesn't seem too crazy yet.

Comment 41

[Ed Morley [:emorley]]

My pass was more of a f+

Comment 42

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2151/#review1413

Comment 43

[Steve Fink [:sfink] [:s:]]

Ok, I've reviewed a bunch of pieces here. I'm at a bit of a loss to figure out what else I should look at, or how far along these changes are. I didn't see anything that made me more comfortable about the profiles[0] thing, at least, so I know at least one of these isn't marked shipit yet.

Comment 44

[Ed Morley [:emorley]]

(In reply to Steve Fink [:sfink, :s:] from comment #43)
> I didn't
> see anything that made me more comfortable about the profiles[0] thing, at
> least, so I know at least one of these isn't marked shipit yet.

https://reviewboard.mozilla.org/r/2147/ makes bzexport use:
https://hg.mozilla.org/hgcustom/version-control-tools/file/f7ba7ce1adda/pylib/mozhg/mozhg/auth.py#l62

Comment 45

[Steve Fink [:sfink] [:s:]]

(In reply to Ed Morley [:edmorley] from comment #44)
> (In reply to Steve Fink [:sfink, :s:] from comment #43)
> > I didn't
> > see anything that made me more comfortable about the profiles[0] thing, at
> > least, so I know at least one of these isn't marked shipit yet.
> 
> https://reviewboard.mozilla.org/r/2147/ makes bzexport use:
> https://hg.mozilla.org/hgcustom/version-control-tools/file/f7ba7ce1adda/
> pylib/mozhg/mozhg/auth.py#l62

Ah, so it does. Thanks.

Comment 46

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2143/#review1419

Ok, the badness goes away later.

Comment 47

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2127/#review1421

Ship It!

Comment 48

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2145/#review1427

Sorry. That wasn't my point, but it turns out I was blind. It does indeed switch to the new name in this very patch. Ignore me.

Comment 49

[Gregory Szorc [:gps]]

/r/2129 - tests: dump more bug state
/r/2261 - bugzilla: do not use Bugsy for fetching bugs
/r/2263 - testing: add cc, blocks, and depends_on to bug dump
/r/2131 - bzexport: test that uploading a replacement patch works
/r/2133 - bzexport: add test for reviewer selection
/r/2135 - bzexport: partially implement reviewer switching test
/r/2137 - bzexport: enable loading modules from version-control-tools
/r/2139 - bzexport: use win_get_folder_path from mozhg.auth
/r/2141 - bzexport: use profile path finding from mozhg.auth
/r/2143 - bzexport: use profiles loading code from mozhg.auth
/r/2145 - bzexport: use mozhg.auth for finding cookies from profile
/r/2147 - bzexport: use mozhg.auth.getbugzillaauth
/r/2149 - bzexport: support for using a requests.Session for making requests
/r/2151 - bzexport: use REST API to obsolete attachments (bug 1033394)
/r/2265 - bzexport: deprecate bzexport.username and bzexport.password
/r/2267 - bzexport: test for assigning a bug
/r/2269 - bzexport: test for CC list on newbug
/r/2271 - bzexport: test for setting bug dependencies with newbug
/r/2273 - bzexport: add an API for performing REST requests
/r/2275 - bzexport: use REST for creating bugs
/r/2277 - bzexport: add test for reviewing feedback
/r/2279 - bzexport: fetch attachments via REST API
/r/2281 - bzexport: use REST API for attachment creation
/r/2283 - bzexport: use REST API for getting and updating bugs
/r/2285 - bzexport: use REST to obtain information about a user
/r/2287 - bzexport: search for users using REST API

Pull down these commits:

hg pull review -r 1fae1cb2295b07593d9491adb5d746f7a01891eb

Comment 50

[Gregory Szorc [:gps]]

https://reviewboard.mozilla.org/r/2127/#review1431

With this latest patch series, I've killed all uses of bzAPI except `bzapi/configuration`. That one is slightly more non-trivial than the others because the API doesn't map nicely to what's in the REST API AFAICT. And, some semi-complicated caching code in bzexport doesn't hurt either.

I still need to write cookie tests for this. I wouldn't be surprised if cookie auth were somehow busted.

I'm feeling pretty good about the test thoroughness. If we roll this out and it breaks something, it should be relatively easy to code up a test and code a fix to what we know the local Bugzilla is willing to accept.

Comment 51

[Gregory Szorc [:gps]]

https://reviewboard.mozilla.org/r/2143/#review1435

There is a custom sort function. The default profile is always first. So, behavior is preserved.

One thing that may bite us is when mozhg.auth attempts to read *all* profiles and returns cookies from the first one it finds.

I am very receptive to ideas to improve the configurability of mozhg.auth to work in more use cases. As I said, the module is very well tested, so we just need to write the code to appease people.

Comment 52

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2271/#review1441

Ship It!

Comment 53

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2269/#review1443

Good spot, thank you :-)

Comment 54

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2267/#review1445

Ship It!

Comment 55

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2265/#review1447

::: hgext/bzexport/__init__.py
(Diff revision 1)
> +            'use bugzilla.password instead)\n')

Perhaps we could add "Or alternatively use the cookie authentication options instead." to promote moving away from password. (I'd suggest many people don't realise cookie support now exists).

Comment 56

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2263/#review1449

Ship It!

Comment 57

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2285/#review1451

Ship It!

Comment 58

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2273/#review1453

Ship It!

Comment 59

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2275/#review1455

Ship It!

Comment 60

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2277/#review1457

Ship It!

Comment 61

[Ed Morley [:emorley]]

https://reviewboard.mozilla.org/r/2279/#review1459

Ship It!

Comment 62

[Ed Morley [:emorley]]

Comment on attachment 8545777 [details]
MozReview Request: bz://1033394/gps

Take this as somewhere between an f+ and an r+, it's probably best if the others take a look too. Thank you for all the cleanup! :-)

Comment 63

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2265/#review1601

Ship It!

Comment 64

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2279/#review1603

::: hgext/bzexport/__init__.py
(Diff revision 1)
>      try:
> -        bug = json.load(urlopen(ui, req))
> -    except Exception, e:
> -        raise util.Abort(_("Could not load info for bug %s: %s") % (bug, str(e)))
> +        bugs = bz.get_attachments(auth, bugid)
> +    except Exception as e:
> +        raise util.Abort(e.message)
> +
> +    attachments = bugs['bugs'][bugid]

Huh?
bug = get_attachments()?
attachments = bugs['bugs'][bugid]?

Assuming this code is correct, then maybe

bug_attachments = get_attachments()
attachments = bug_attachments['bugs'][bugid]

Comment 65

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2279/#review1605

Ship It!

Comment 66

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2127/#review1599

::: hgext/bzexport/__init__.py
(Diff revision 2)
> +    aid = result['attachments'].keys()[0]

I'd prefer a longer name than 'aid'. Took me a second to figure out it was attachment_id. attachment_id, attach_id, att_id would all be fine.

Comment 67

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2275/#review1607

Ship It!

Comment 68

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2273/#review1611

Ship It!

Comment 69

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2281/#review1613

Ship It!

Comment 70

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2283/#review1615

Ship It!

Comment 71

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2285/#review1617

Ship It!

Comment 72

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2287/#review1619

Ship It!

Comment 73

[Steve Fink [:sfink] [:s:]]

(In reply to Ed Morley [:edmorley] from comment #62)
> Comment on attachment 8545777 [details]
> MozReview Request: bz://1033394/gps
> 
> Take this as somewhere between an f+ and an r+, it's probably best if the
> others take a look too. Thank you for all the cleanup! :-)

Personally, I think edmorley counts as an r+, there's nothing my eyeballs can really add above his, other than just being an additional set.

(And can I request that RB *not* post individual Ship It! comments? Buffer them up and post them all as one comment, maybe, or make the status more visible in RB and don't bother posting them at all. Even for people who want bugzilla to track everything, I don't think it helps.)

Comment 74

[Steve Fink [:sfink] [:s:]]

Comment on attachment 8545777 [details]
MozReview Request: bz://1033394/gps

I think I looked at everything of interest (so far? I don't know if there's more that'll happen here.)

Comment 75

[Mark Côté [:mcote]]

(In reply to Steve Fink [:sfink, :s:] from comment #73)
> (And can I request that RB *not* post individual Ship It! comments? Buffer
> them up and post them all as one comment, maybe, or make the status more
> visible in RB and don't bother posting them at all. Even for people who want
> bugzilla to track everything, I don't think it helps.)

Yeah, we are almost certainly going to reduce the amount of mirroring from Review Board to Bugzilla and make the former the source of truth for anything reviewed by MozReview.  Furthermore, anything mirrored will probably have bugmail suppressed in favour of Review Board emails (although admittedly we might have to tweak those to avoid the same spam coming from a different source).

Comment 76

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2127/#review1621

Ship It!

Comment 77

[Steve Fink [:sfink] [:s:]]

https://reviewboard.mozilla.org/r/2147/#review1623

Ship It!

Comment 78

[Gregory Szorc [:gps]]

https://reviewboard.mozilla.org/r/2265/#review1633

> Perhaps we could add "Or alternatively use the cookie authentication options instead." to promote moving away from password. (I'd suggest many people don't realise cookie support now exists).

I'll amend the message before landing to say something like "or log into Bugzilla using Firefox".

Comment 79

[Gregory Szorc [:gps]]

https://hg.mozilla.org/hgcustom/version-control-tools/rev/b3c1fc01d493

Comment 80

[Steven MacLeod [:smacleod]]

https://reviewboard.mozilla.org/r/2129/#review1635

Ship It!

Comment 81

[Steven MacLeod [:smacleod]]

Comment on attachment 8545777 [details]
MozReview Request: bz://1033394/gps

I know you just landed this, but I just finished looking over the parts you asked r? to me. LGTM.

Comment 82

[Gregory Szorc [:gps]]

Everything I've written has been reviewed and landed.

As mentioned in comment 50, we now use REST for everything except bzapi/configuration. I'll file a follow-up to track that.

Comment 85

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use REST API for getting and updating bugs

Comment 86

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use REST for creating bugs

Comment 87

[Gregory Szorc [:gps]]

Attachment: MozReview Request: testing: add cc, blocks, and depends_on to bug dump

Comment 88

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: deprecate bzexport.username and bzexport.password

Comment 89

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: support for using a requests.Session for making requests

Comment 90

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use win_get_folder_path from mozhg.auth

Comment 91

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: test for CC list on newbug

Comment 92

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: test for setting bug dependencies with newbug

Comment 93

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use mozhg.auth for finding cookies from profile

Comment 94

[Gregory Szorc [:gps]]

Attachment: MozReview Request: tests: dump more bug state

Comment 95

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: test that uploading a replacement patch works

Comment 96

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: add test for reviewer selection

Comment 97

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: partially implement reviewer switching test

Comment 98

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: enable loading modules from version-control-tools

Comment 99

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: test for assigning a bug

Comment 100

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use profile path finding from mozhg.auth

Comment 101

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use profiles loading code from mozhg.auth

Comment 102

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: add an API for performing REST requests

Comment 103

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use mozhg.auth.getbugzillaauth

Comment 104

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: add test for reviewing feedback

Comment 105

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: fetch attachments via REST API

Comment 106

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use REST API for attachment creation

Comment 107

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use REST API to obsolete attachments (bug 1033394)

Comment 108

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: use REST to obtain information about a user

Comment 109

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bzexport: search for users using REST API

Comment 110

[Gregory Szorc [:gps]]

Attachment: MozReview Request: bugzilla: do not use Bugsy for fetching bugs