Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hi Sebastian, Hmm, that's unfortunate. We have recently found an edge case ourselves involving the new algorithm, but need to collect more samples. Would it be possible to collect some information from you privately about the diff and its contents? We're happy to sign an NDA. Christian On Sun, Jun 11, 2023 at 3:19 PM sebun...@gmail.com wrote: > This is still an issue in 5.0.4. > > On Friday, July 14, 2017 at 12:29:53 PM UTC+12 Sebastian Unger wrote: > >> We've just come across this (or something very similar) in 2.5.13.1. >> >> >>- Is this expected (i.e. you know the bug is still there) or >>something new (i.e. you fixed the bug and this is something else)? >>- Is there a bug tracker issue somewhere that I can subscribe to to >>get updates regarding this? >>- Are the circumstances under which this happens known so I can >>inform my users what to look out for? >>- Is there any additional testing or debugging I can help with to get >>this resolved? >> >> Cheers, >> Seb >> >> >> >> On Wednesday, 12 July 2017 05:46:52 UTC+12, Will Jacobs wrote: >>> >>> Awesome, thanks again! >>> >>> On Tuesday, July 11, 2017 at 11:31:07 AM UTC-5, Christian Hammond wrote: >>>> >>>> Hey Will, >>>> >>>> It's a bug in 2.5.x and later 2.0.x releases. It wasn't present in >>>> 1.6.x or 1.7.x. >>>> >>>> Christian >>>> >>>> On Tue, Jul 11, 2017 at 9:10 AM, Will Jacobs >>>> wrote: >>>> >>>>> Hey Christian - >>>>> >>>>> Thanks as usual for the response! Just a quick question - how long has >>>>> this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am >>>>> seeing this as well, but I'm trying to understand if it was also a bug in >>>>> 1.6.14 which we never noticed or never exhibited. >>>>> >>>>> Thanks, >>>>> -Will >>>>> >>>>> On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: >>>>>> >>>>>> Thanks, Peter. There's still bugs here, and I've spent a lot of time >>>>>> reworking the algorithms to fix these. I'm not ready to ship any of that >>>>>> code yet, though. Needs further tweaking and testing. Having test cases >>>>>> like these really help with that, and I'll add the files to my regression >>>>>> tests. >>>>>> >>>>>> Christian >>>>>> >>>>>> On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson >>>>>> wrote: >>>>>> >>>>>>> Hi Chris, >>>>>>> >>>>>>> I've now replicated this with a very simple example. >>>>>>> >>>>>>> The original file is attached as LICENSE. >>>>>>> >>>>>>> I added a char and committed the change reflected in rb40.patch and >>>>>>> created the review request (that file is created via the download patch >>>>>>> link) >>>>>>> >>>>>>> I then committed a change removing the line starting with the word >>>>>>> LIABILITY and updated the review request, the full diff of which is the >>>>>>> file rb40_v2.patch (again created via the download patch link). >>>>>>> >>>>>>> To define the issue again: >>>>>>> orig - v2 Shows the correct diff, showing both the char addition and >>>>>>> line removal >>>>>>> orig - v1 Shows the correct diff, showing just the line change >>>>>>> v1 - v2Is nonsense, the diff claims "This file contains only >>>>>>> whitespace changes" and shows the last 5 lines of the rev1 file vs the >>>>>>> last >>>>>>> four of the rev2 file. It does not highlight the missing lines. >>>>>>> >>>>>>> I understand this is an old thread, if I don't hear back in a couple >>>>>>> of days I'll raise it afresh. >>>>>>> >>>>>>> Many thanks, >>>>>>> Peter >>>>>>> >>>>>>> >>>>>>> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>>>>>>> >>>>>>>> Hi Chris, >&g
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
This is still an issue in 5.0.4. On Friday, July 14, 2017 at 12:29:53 PM UTC+12 Sebastian Unger wrote: > We've just come across this (or something very similar) in 2.5.13.1. > > >- Is this expected (i.e. you know the bug is still there) or something >new (i.e. you fixed the bug and this is something else)? >- Is there a bug tracker issue somewhere that I can subscribe to to >get updates regarding this? >- Are the circumstances under which this happens known so I can inform >my users what to look out for? >- Is there any additional testing or debugging I can help with to get >this resolved? > > Cheers, > Seb > > > > On Wednesday, 12 July 2017 05:46:52 UTC+12, Will Jacobs wrote: >> >> Awesome, thanks again! >> >> On Tuesday, July 11, 2017 at 11:31:07 AM UTC-5, Christian Hammond wrote: >>> >>> Hey Will, >>> >>> It's a bug in 2.5.x and later 2.0.x releases. It wasn't present in 1.6.x >>> or 1.7.x. >>> >>> Christian >>> >>> On Tue, Jul 11, 2017 at 9:10 AM, Will Jacobs wrote: >>> >>>> Hey Christian - >>>> >>>> Thanks as usual for the response! Just a quick question - how long has >>>> this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am >>>> seeing this as well, but I'm trying to understand if it was also a bug in >>>> 1.6.14 which we never noticed or never exhibited. >>>> >>>> Thanks, >>>> -Will >>>> >>>> On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: >>>>> >>>>> Thanks, Peter. There's still bugs here, and I've spent a lot of time >>>>> reworking the algorithms to fix these. I'm not ready to ship any of that >>>>> code yet, though. Needs further tweaking and testing. Having test cases >>>>> like these really help with that, and I'll add the files to my regression >>>>> tests. >>>>> >>>>> Christian >>>>> >>>>> On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson >>>>> wrote: >>>>> >>>>>> Hi Chris, >>>>>> >>>>>> I've now replicated this with a very simple example. >>>>>> >>>>>> The original file is attached as LICENSE. >>>>>> >>>>>> I added a char and committed the change reflected in rb40.patch and >>>>>> created the review request (that file is created via the download patch >>>>>> link) >>>>>> >>>>>> I then committed a change removing the line starting with the word >>>>>> LIABILITY and updated the review request, the full diff of which is the >>>>>> file rb40_v2.patch (again created via the download patch link). >>>>>> >>>>>> To define the issue again: >>>>>> orig - v2 Shows the correct diff, showing both the char addition and >>>>>> line removal >>>>>> orig - v1 Shows the correct diff, showing just the line change >>>>>> v1 - v2Is nonsense, the diff claims "This file contains only >>>>>> whitespace changes" and shows the last 5 lines of the rev1 file vs the >>>>>> last >>>>>> four of the rev2 file. It does not highlight the missing lines. >>>>>> >>>>>> I understand this is an old thread, if I don't hear back in a couple >>>>>> of days I'll raise it afresh. >>>>>> >>>>>> Many thanks, >>>>>> Peter >>>>>> >>>>>> >>>>>> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>>>>>> >>>>>>> Hi Chris, >>>>>>> >>>>>>> I just came across this same bug in 2.5.12 exactly as described by >>>>>>> Paul so I'm presuming it wasn't fixed. >>>>>>> >>>>>>> What files do you need to debug this? I'll see if I can reproduce >>>>>>> this without distributing our actual codebase. >>>>>>> >>>>>>> Thanks, >>>>>>> Peter >>>>>>> >>>>>>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>>>>>> >>>>>>>> Hi Paul, >>>>>>>> >>&
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
We've just come across this (or something very similar) in 2.5.13.1. - Is this expected (i.e. you know the bug is still there) or something new (i.e. you fixed the bug and this is something else)? - Is there a bug tracker issue somewhere that I can subscribe to to get updates regarding this? - Are the circumstances under which this happens known so I can inform my users what to look out for? - Is there any additional testing or debugging I can help with to get this resolved? Cheers, Seb On Wednesday, 12 July 2017 05:46:52 UTC+12, Will Jacobs wrote: > > Awesome, thanks again! > > On Tuesday, July 11, 2017 at 11:31:07 AM UTC-5, Christian Hammond wrote: >> >> Hey Will, >> >> It's a bug in 2.5.x and later 2.0.x releases. It wasn't present in 1.6.x >> or 1.7.x. >> >> Christian >> >> On Tue, Jul 11, 2017 at 9:10 AM, Will Jacobs <will@gmail.com> wrote: >> >>> Hey Christian - >>> >>> Thanks as usual for the response! Just a quick question - how long has >>> this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am >>> seeing this as well, but I'm trying to understand if it was also a bug in >>> 1.6.14 which we never noticed or never exhibited. >>> >>> Thanks, >>> -Will >>> >>> On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: >>>> >>>> Thanks, Peter. There's still bugs here, and I've spent a lot of time >>>> reworking the algorithms to fix these. I'm not ready to ship any of that >>>> code yet, though. Needs further tweaking and testing. Having test cases >>>> like these really help with that, and I'll add the files to my regression >>>> tests. >>>> >>>> Christian >>>> >>>> On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb@gmail.com> >>>> wrote: >>>> >>>>> Hi Chris, >>>>> >>>>> I've now replicated this with a very simple example. >>>>> >>>>> The original file is attached as LICENSE. >>>>> >>>>> I added a char and committed the change reflected in rb40.patch and >>>>> created the review request (that file is created via the download patch >>>>> link) >>>>> >>>>> I then committed a change removing the line starting with the word >>>>> LIABILITY and updated the review request, the full diff of which is the >>>>> file rb40_v2.patch (again created via the download patch link). >>>>> >>>>> To define the issue again: >>>>> orig - v2 Shows the correct diff, showing both the char addition and >>>>> line removal >>>>> orig - v1 Shows the correct diff, showing just the line change >>>>> v1 - v2Is nonsense, the diff claims "This file contains only >>>>> whitespace changes" and shows the last 5 lines of the rev1 file vs the >>>>> last >>>>> four of the rev2 file. It does not highlight the missing lines. >>>>> >>>>> I understand this is an old thread, if I don't hear back in a couple >>>>> of days I'll raise it afresh. >>>>> >>>>> Many thanks, >>>>> Peter >>>>> >>>>> >>>>> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>>>>> >>>>>> Hi Chris, >>>>>> >>>>>> I just came across this same bug in 2.5.12 exactly as described by >>>>>> Paul so I'm presuming it wasn't fixed. >>>>>> >>>>>> What files do you need to debug this? I'll see if I can reproduce >>>>>> this without distributing our actual codebase. >>>>>> >>>>>> Thanks, >>>>>> Peter >>>>>> >>>>>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>>>>> >>>>>>> Hi Paul, >>>>>>> >>>>>>> We recently fixed a bug like this, but perhaps there's another issue >>>>>>> somewhere. >>>>>>> >>>>>>> In order to diagnose this, I'll need to have original copies of the >>>>>>> affected files, as well as both diffs. I won't be able to diagnose >>>>>>> without >>>>>>> those, unfortunately. >>>>>>> &
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Awesome, thanks again! On Tuesday, July 11, 2017 at 11:31:07 AM UTC-5, Christian Hammond wrote: > > Hey Will, > > It's a bug in 2.5.x and later 2.0.x releases. It wasn't present in 1.6.x > or 1.7.x. > > Christian > > On Tue, Jul 11, 2017 at 9:10 AM, Will Jacobs <will@gmail.com > > wrote: > >> Hey Christian - >> >> Thanks as usual for the response! Just a quick question - how long has >> this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am >> seeing this as well, but I'm trying to understand if it was also a bug in >> 1.6.14 which we never noticed or never exhibited. >> >> Thanks, >> -Will >> >> On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: >>> >>> Thanks, Peter. There's still bugs here, and I've spent a lot of time >>> reworking the algorithms to fix these. I'm not ready to ship any of that >>> code yet, though. Needs further tweaking and testing. Having test cases >>> like these really help with that, and I'll add the files to my regression >>> tests. >>> >>> Christian >>> >>> On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb@gmail.com> >>> wrote: >>> >>>> Hi Chris, >>>> >>>> I've now replicated this with a very simple example. >>>> >>>> The original file is attached as LICENSE. >>>> >>>> I added a char and committed the change reflected in rb40.patch and >>>> created the review request (that file is created via the download patch >>>> link) >>>> >>>> I then committed a change removing the line starting with the word >>>> LIABILITY and updated the review request, the full diff of which is the >>>> file rb40_v2.patch (again created via the download patch link). >>>> >>>> To define the issue again: >>>> orig - v2 Shows the correct diff, showing both the char addition and >>>> line removal >>>> orig - v1 Shows the correct diff, showing just the line change >>>> v1 - v2Is nonsense, the diff claims "This file contains only >>>> whitespace changes" and shows the last 5 lines of the rev1 file vs the >>>> last >>>> four of the rev2 file. It does not highlight the missing lines. >>>> >>>> I understand this is an old thread, if I don't hear back in a couple of >>>> days I'll raise it afresh. >>>> >>>> Many thanks, >>>> Peter >>>> >>>> >>>> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>>>> >>>>> Hi Chris, >>>>> >>>>> I just came across this same bug in 2.5.12 exactly as described by >>>>> Paul so I'm presuming it wasn't fixed. >>>>> >>>>> What files do you need to debug this? I'll see if I can reproduce this >>>>> without distributing our actual codebase. >>>>> >>>>> Thanks, >>>>> Peter >>>>> >>>>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>>>> >>>>>> Hi Paul, >>>>>> >>>>>> We recently fixed a bug like this, but perhaps there's another issue >>>>>> somewhere. >>>>>> >>>>>> In order to diagnose this, I'll need to have original copies of the >>>>>> affected files, as well as both diffs. I won't be able to diagnose >>>>>> without >>>>>> those, unfortunately. >>>>>> >>>>>> Christian >>>>>> >>>>>> -- >>>>>> Christian Hammond - chri...@beanbaginc.com >>>>>> Review Board - https://www.reviewboard.org >>>>>> Beanbag, Inc. - https://www.beanbaginc.com >>>>>> >>>>>> -Original Message- >>>>>> From: Paul Fee <paul@gmail.com> >>>>>> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >>>>>> Date: June 30, 2015 at 4:02:10 AM >>>>>> To: revie...@googlegroups.com <revie...@googlegroups.com>> >>>>>> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >>>>>> >>>>>> > Hi all, >>>>>> > >>>>>> > I'm using ReviewBoard 2.0.17 and see unexpec
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hey Will, It's a bug in 2.5.x and later 2.0.x releases. It wasn't present in 1.6.x or 1.7.x. Christian On Tue, Jul 11, 2017 at 9:10 AM, Will Jacobs <will.jac...@gmail.com> wrote: > Hey Christian - > > Thanks as usual for the response! Just a quick question - how long has > this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am > seeing this as well, but I'm trying to understand if it was also a bug in > 1.6.14 which we never noticed or never exhibited. > > Thanks, > -Will > > On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: >> >> Thanks, Peter. There's still bugs here, and I've spent a lot of time >> reworking the algorithms to fix these. I'm not ready to ship any of that >> code yet, though. Needs further tweaking and testing. Having test cases >> like these really help with that, and I'll add the files to my regression >> tests. >> >> Christian >> >> On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb@gmail.com> >> wrote: >> >>> Hi Chris, >>> >>> I've now replicated this with a very simple example. >>> >>> The original file is attached as LICENSE. >>> >>> I added a char and committed the change reflected in rb40.patch and >>> created the review request (that file is created via the download patch >>> link) >>> >>> I then committed a change removing the line starting with the word >>> LIABILITY and updated the review request, the full diff of which is the >>> file rb40_v2.patch (again created via the download patch link). >>> >>> To define the issue again: >>> orig - v2 Shows the correct diff, showing both the char addition and >>> line removal >>> orig - v1 Shows the correct diff, showing just the line change >>> v1 - v2Is nonsense, the diff claims "This file contains only >>> whitespace changes" and shows the last 5 lines of the rev1 file vs the last >>> four of the rev2 file. It does not highlight the missing lines. >>> >>> I understand this is an old thread, if I don't hear back in a couple of >>> days I'll raise it afresh. >>> >>> Many thanks, >>> Peter >>> >>> >>> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>>> >>>> Hi Chris, >>>> >>>> I just came across this same bug in 2.5.12 exactly as described by Paul >>>> so I'm presuming it wasn't fixed. >>>> >>>> What files do you need to debug this? I'll see if I can reproduce this >>>> without distributing our actual codebase. >>>> >>>> Thanks, >>>> Peter >>>> >>>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>>> >>>>> Hi Paul, >>>>> >>>>> We recently fixed a bug like this, but perhaps there's another issue >>>>> somewhere. >>>>> >>>>> In order to diagnose this, I'll need to have original copies of the >>>>> affected files, as well as both diffs. I won't be able to diagnose without >>>>> those, unfortunately. >>>>> >>>>> Christian >>>>> >>>>> -- >>>>> Christian Hammond - chri...@beanbaginc.com >>>>> Review Board - https://www.reviewboard.org >>>>> Beanbag, Inc. - https://www.beanbaginc.com >>>>> >>>>> -Original Message- >>>>> From: Paul Fee <paul@gmail.com> >>>>> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >>>>> Date: June 30, 2015 at 4:02:10 AM >>>>> To: revie...@googlegroups.com <revie...@googlegroups.com>> >>>>> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >>>>> >>>>> > Hi all, >>>>> > >>>>> > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the >>>>> diff >>>>> > viewer. >>>>> > >>>>> > Steps to reproduce: >>>>> > >>>>> > 1. Change a few files (I'm using SVN) >>>>> > 2. rbt post >>>>> > 3. Publish review >>>>> > 4. Delete a group of lines from one of the files already changed. >>>>> > 5. rbt post -r >>>>> > 6. Publish review >>>>> > 7. Review entire diff: http://reviewboard/r/9581/di
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hey Christian - Thanks as usual for the response! Just a quick question - how long has this bug been around? I recently upgraded from 1.6.14 to 2.5.12, and am seeing this as well, but I'm trying to understand if it was also a bug in 1.6.14 which we never noticed or never exhibited. Thanks, -Will On Thursday, July 6, 2017 at 6:14:07 AM UTC-5, Christian Hammond wrote: > > Thanks, Peter. There's still bugs here, and I've spent a lot of time > reworking the algorithms to fix these. I'm not ready to ship any of that > code yet, though. Needs further tweaking and testing. Having test cases > like these really help with that, and I'll add the files to my regression > tests. > > Christian > > On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb@gmail.com > > wrote: > >> Hi Chris, >> >> I've now replicated this with a very simple example. >> >> The original file is attached as LICENSE. >> >> I added a char and committed the change reflected in rb40.patch and >> created the review request (that file is created via the download patch >> link) >> >> I then committed a change removing the line starting with the word >> LIABILITY and updated the review request, the full diff of which is the >> file rb40_v2.patch (again created via the download patch link). >> >> To define the issue again: >> orig - v2 Shows the correct diff, showing both the char addition and line >> removal >> orig - v1 Shows the correct diff, showing just the line change >> v1 - v2Is nonsense, the diff claims "This file contains only >> whitespace changes" and shows the last 5 lines of the rev1 file vs the last >> four of the rev2 file. It does not highlight the missing lines. >> >> I understand this is an old thread, if I don't hear back in a couple of >> days I'll raise it afresh. >> >> Many thanks, >> Peter >> >> >> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>> >>> Hi Chris, >>> >>> I just came across this same bug in 2.5.12 exactly as described by Paul >>> so I'm presuming it wasn't fixed. >>> >>> What files do you need to debug this? I'll see if I can reproduce this >>> without distributing our actual codebase. >>> >>> Thanks, >>> Peter >>> >>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>> >>>> Hi Paul, >>>> >>>> We recently fixed a bug like this, but perhaps there's another issue >>>> somewhere. >>>> >>>> In order to diagnose this, I'll need to have original copies of the >>>> affected files, as well as both diffs. I won't be able to diagnose without >>>> those, unfortunately. >>>> >>>> Christian >>>> >>>> -- >>>> Christian Hammond - chri...@beanbaginc.com >>>> Review Board - https://www.reviewboard.org >>>> Beanbag, Inc. - https://www.beanbaginc.com >>>> >>>> -Original Message- >>>> From: Paul Fee <paul@gmail.com> >>>> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >>>> Date: June 30, 2015 at 4:02:10 AM >>>> To: revie...@googlegroups.com <revie...@googlegroups.com>> >>>> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >>>> >>>> > Hi all, >>>> > >>>> > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff >>>> > viewer. >>>> > >>>> > Steps to reproduce: >>>> > >>>> > 1. Change a few files (I'm using SVN) >>>> > 2. rbt post >>>> > 3. Publish review >>>> > 4. Delete a group of lines from one of the files already changed. >>>> > 5. rbt post -r >>>> > 6. Publish review >>>> > 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: >>>> PASS >>>> > 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - >>>> Result: PASS >>>> > 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - >>>> Result: >>>> > FAIL >>>> > >>>> > In the second interdiff, RB states that the file contains only >>>> whitespace >>>> > changes, this is incorrect as lines have been deleted. >>>> > >>>> > Expanding the entire file, the co
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Excellent, that sounds good Christian, Thanks for the reply, Peter On Thursday, 6 July 2017 12:14:07 UTC+1, Christian Hammond wrote: > > Thanks, Peter. There's still bugs here, and I've spent a lot of time > reworking the algorithms to fix these. I'm not ready to ship any of that > code yet, though. Needs further tweaking and testing. Having test cases > like these really help with that, and I'll add the files to my regression > tests. > > Christian > > On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb@gmail.com > > wrote: > >> Hi Chris, >> >> I've now replicated this with a very simple example. >> >> The original file is attached as LICENSE. >> >> I added a char and committed the change reflected in rb40.patch and >> created the review request (that file is created via the download patch >> link) >> >> I then committed a change removing the line starting with the word >> LIABILITY and updated the review request, the full diff of which is the >> file rb40_v2.patch (again created via the download patch link). >> >> To define the issue again: >> orig - v2 Shows the correct diff, showing both the char addition and line >> removal >> orig - v1 Shows the correct diff, showing just the line change >> v1 - v2Is nonsense, the diff claims "This file contains only >> whitespace changes" and shows the last 5 lines of the rev1 file vs the last >> four of the rev2 file. It does not highlight the missing lines. >> >> I understand this is an old thread, if I don't hear back in a couple of >> days I'll raise it afresh. >> >> Many thanks, >> Peter >> >> >> On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >>> >>> Hi Chris, >>> >>> I just came across this same bug in 2.5.12 exactly as described by Paul >>> so I'm presuming it wasn't fixed. >>> >>> What files do you need to debug this? I'll see if I can reproduce this >>> without distributing our actual codebase. >>> >>> Thanks, >>> Peter >>> >>> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>>> >>>> Hi Paul, >>>> >>>> We recently fixed a bug like this, but perhaps there's another issue >>>> somewhere. >>>> >>>> In order to diagnose this, I'll need to have original copies of the >>>> affected files, as well as both diffs. I won't be able to diagnose without >>>> those, unfortunately. >>>> >>>> Christian >>>> >>>> -- >>>> Christian Hammond - chri...@beanbaginc.com >>>> Review Board - https://www.reviewboard.org >>>> Beanbag, Inc. - https://www.beanbaginc.com >>>> >>>> -Original Message- >>>> From: Paul Fee <paul@gmail.com> >>>> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >>>> Date: June 30, 2015 at 4:02:10 AM >>>> To: revie...@googlegroups.com <revie...@googlegroups.com>> >>>> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >>>> >>>> > Hi all, >>>> > >>>> > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff >>>> > viewer. >>>> > >>>> > Steps to reproduce: >>>> > >>>> > 1. Change a few files (I'm using SVN) >>>> > 2. rbt post >>>> > 3. Publish review >>>> > 4. Delete a group of lines from one of the files already changed. >>>> > 5. rbt post -r >>>> > 6. Publish review >>>> > 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: >>>> PASS >>>> > 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - >>>> Result: PASS >>>> > 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - >>>> Result: >>>> > FAIL >>>> > >>>> > In the second interdiff, RB states that the file contains only >>>> whitespace >>>> > changes, this is incorrect as lines have been deleted. >>>> > >>>> > Expanding the entire file, the contain on the left and right diff >>>> panels is >>>> > correct, I can see that the deleted lines have gone, however they're >>>> not >>>> > highlighted in red as expected. >&g
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Thanks, Peter. There's still bugs here, and I've spent a lot of time reworking the algorithms to fix these. I'm not ready to ship any of that code yet, though. Needs further tweaking and testing. Having test cases like these really help with that, and I'll add the files to my regression tests. Christian On Wed, Jul 5, 2017 at 9:20 AM, Peter Hodgson <peterb.hodg...@gmail.com> wrote: > Hi Chris, > > I've now replicated this with a very simple example. > > The original file is attached as LICENSE. > > I added a char and committed the change reflected in rb40.patch and > created the review request (that file is created via the download patch > link) > > I then committed a change removing the line starting with the word > LIABILITY and updated the review request, the full diff of which is the > file rb40_v2.patch (again created via the download patch link). > > To define the issue again: > orig - v2 Shows the correct diff, showing both the char addition and line > removal > orig - v1 Shows the correct diff, showing just the line change > v1 - v2Is nonsense, the diff claims "This file contains only > whitespace changes" and shows the last 5 lines of the rev1 file vs the last > four of the rev2 file. It does not highlight the missing lines. > > I understand this is an old thread, if I don't hear back in a couple of > days I'll raise it afresh. > > Many thanks, > Peter > > > On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: >> >> Hi Chris, >> >> I just came across this same bug in 2.5.12 exactly as described by Paul >> so I'm presuming it wasn't fixed. >> >> What files do you need to debug this? I'll see if I can reproduce this >> without distributing our actual codebase. >> >> Thanks, >> Peter >> >> On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >>> >>> Hi Paul, >>> >>> We recently fixed a bug like this, but perhaps there's another issue >>> somewhere. >>> >>> In order to diagnose this, I'll need to have original copies of the >>> affected files, as well as both diffs. I won't be able to diagnose without >>> those, unfortunately. >>> >>> Christian >>> >>> -- >>> Christian Hammond - chri...@beanbaginc.com >>> Review Board - https://www.reviewboard.org >>> Beanbag, Inc. - https://www.beanbaginc.com >>> >>> -Original Message- >>> From: Paul Fee <paul@gmail.com> >>> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >>> Date: June 30, 2015 at 4:02:10 AM >>> To: revie...@googlegroups.com <revie...@googlegroups.com>> >>> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >>> >>> > Hi all, >>> > >>> > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff >>> > viewer. >>> > >>> > Steps to reproduce: >>> > >>> > 1. Change a few files (I'm using SVN) >>> > 2. rbt post >>> > 3. Publish review >>> > 4. Delete a group of lines from one of the files already changed. >>> > 5. rbt post -r >>> > 6. Publish review >>> > 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: >>> PASS >>> > 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - >>> Result: PASS >>> > 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - >>> Result: >>> > FAIL >>> > >>> > In the second interdiff, RB states that the file contains only >>> whitespace >>> > changes, this is incorrect as lines have been deleted. >>> > >>> > Expanding the entire file, the contain on the left and right diff >>> panels is >>> > correct, I can see that the deleted lines have gone, however they're >>> not >>> > highlighted in red as expected. >>> > >>> > I don't think this is related to caching as the following steps had no >>> > effect, the second interdiff consistently shows the same result. >>> > >>> > * systemctl restart memcached >>> > * systemctl restart httpd >>> > * View second interdiff with different browsers (Firefox and >>> Chromium), >>> > both show same results, hence not a browser cache issue. >>> > >>> > I'm running ReviewBoard on CentOS7 using EPEL packages. >>> > >>> > Let me know if you need more information to help recreate or
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hi Chris, I've now replicated this with a very simple example. The original file is attached as LICENSE. I added a char and committed the change reflected in rb40.patch and created the review request (that file is created via the download patch link) I then committed a change removing the line starting with the word LIABILITY and updated the review request, the full diff of which is the file rb40_v2.patch (again created via the download patch link). To define the issue again: orig - v2 Shows the correct diff, showing both the char addition and line removal orig - v1 Shows the correct diff, showing just the line change v1 - v2Is nonsense, the diff claims "This file contains only whitespace changes" and shows the last 5 lines of the rev1 file vs the last four of the rev2 file. It does not highlight the missing lines. I understand this is an old thread, if I don't hear back in a couple of days I'll raise it afresh. Many thanks, Peter On Wednesday, 5 July 2017 16:35:28 UTC+1, Peter Hodgson wrote: > > Hi Chris, > > I just came across this same bug in 2.5.12 exactly as described by Paul so > I'm presuming it wasn't fixed. > > What files do you need to debug this? I'll see if I can reproduce this > without distributing our actual codebase. > > Thanks, > Peter > > On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: >> >> Hi Paul, >> >> We recently fixed a bug like this, but perhaps there's another issue >> somewhere. >> >> In order to diagnose this, I'll need to have original copies of the >> affected files, as well as both diffs. I won't be able to diagnose without >> those, unfortunately. >> >> Christian >> >> -- >> Christian Hammond - chri...@beanbaginc.com >> Review Board - https://www.reviewboard.org >> Beanbag, Inc. - https://www.beanbaginc.com >> >> -Original Message- >> From: Paul Fee <paul@gmail.com> >> Reply: revie...@googlegroups.com <revie...@googlegroups.com>> >> Date: June 30, 2015 at 4:02:10 AM >> To: revie...@googlegroups.com <revie...@googlegroups.com>> >> Subject: RB 2.0.17 - Interdiff fails to highlight deleted code >> >> > Hi all, >> > >> > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff >> > viewer. >> > >> > Steps to reproduce: >> > >> > 1. Change a few files (I'm using SVN) >> > 2. rbt post >> > 3. Publish review >> > 4. Delete a group of lines from one of the files already changed. >> > 5. rbt post -r >> > 6. Publish review >> > 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: >> PASS >> > 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - Result: >> PASS >> > 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - >> Result: >> > FAIL >> > >> > In the second interdiff, RB states that the file contains only >> whitespace >> > changes, this is incorrect as lines have been deleted. >> > >> > Expanding the entire file, the contain on the left and right diff >> panels is >> > correct, I can see that the deleted lines have gone, however they're >> not >> > highlighted in red as expected. >> > >> > I don't think this is related to caching as the following steps had no >> > effect, the second interdiff consistently shows the same result. >> > >> > * systemctl restart memcached >> > * systemctl restart httpd >> > * View second interdiff with different browsers (Firefox and Chromium), >> > both show same results, hence not a browser cache issue. >> > >> > I'm running ReviewBoard on CentOS7 using EPEL packages. >> > >> > Let me know if you need more information to help recreate or fix this >> bug. >> > >> > Thanks, >> > Paul >> > >> > -- >> > Supercharge your Review Board with Power Pack: >> https://www.reviewboard.org/powerpack/ >> > Want us to host Review Board for you? Check out RBCommons: >> https://rbcommons.com/ >> > Happy user? Let us know! https://www.reviewboard.org/users/ >> > --- >> > You received this message because you are subscribed to the Google >> Groups "reviewboard" >> > group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an email to reviewboard...@googlegroups.com. >> > For more options, visit https://g
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hi Chris, I just came across this same bug in 2.5.12 exactly as described by Paul so I'm presuming it wasn't fixed. What files do you need to debug this? I'll see if I can reproduce this without distributing our actual codebase. Thanks, Peter On Tuesday, 30 June 2015 22:37:02 UTC+1, Christian Hammond wrote: > > Hi Paul, > > We recently fixed a bug like this, but perhaps there's another issue > somewhere. > > In order to diagnose this, I'll need to have original copies of the > affected files, as well as both diffs. I won't be able to diagnose without > those, unfortunately. > > Christian > > -- > Christian Hammond - chri...@beanbaginc.com > Review Board - https://www.reviewboard.org > Beanbag, Inc. - https://www.beanbaginc.com > > -Original Message- > From: Paul Fee <paul@gmail.com > > Reply: revie...@googlegroups.com <revie...@googlegroups.com > >> > Date: June 30, 2015 at 4:02:10 AM > To: revie...@googlegroups.com <revie...@googlegroups.com > >> > Subject: RB 2.0.17 - Interdiff fails to highlight deleted code > > > Hi all, > > > > I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff > > viewer. > > > > Steps to reproduce: > > > > 1. Change a few files (I'm using SVN) > > 2. rbt post > > 3. Publish review > > 4. Delete a group of lines from one of the files already changed. > > 5. rbt post -r > > 6. Publish review > > 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: PASS > > 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - Result: > PASS > > 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - > Result: > > FAIL > > > > In the second interdiff, RB states that the file contains only > whitespace > > changes, this is incorrect as lines have been deleted. > > > > Expanding the entire file, the contain on the left and right diff panels > is > > correct, I can see that the deleted lines have gone, however they're not > > highlighted in red as expected. > > > > I don't think this is related to caching as the following steps had no > > effect, the second interdiff consistently shows the same result. > > > > * systemctl restart memcached > > * systemctl restart httpd > > * View second interdiff with different browsers (Firefox and Chromium), > > both show same results, hence not a browser cache issue. > > > > I'm running ReviewBoard on CentOS7 using EPEL packages. > > > > Let me know if you need more information to help recreate or fix this > bug. > > > > Thanks, > > Paul > > > > -- > > Supercharge your Review Board with Power Pack: > https://www.reviewboard.org/powerpack/ > > Want us to host Review Board for you? Check out RBCommons: > https://rbcommons.com/ > > Happy user? Let us know! https://www.reviewboard.org/users/ > > --- > > You received this message because you are subscribed to the Google > Groups "reviewboard" > > group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to reviewboard...@googlegroups.com . > > For more options, visit https://groups.google.com/d/optout. > > > > -- Supercharge your Review Board with Power Pack: https://www.reviewboard.org/powerpack/ Want us to host Review Board for you? Check out RBCommons: https://rbcommons.com/ Happy user? Let us know! https://www.reviewboard.org/users/ --- You received this message because you are subscribed to the Google Groups "reviewboard" group. To unsubscribe from this group and stop receiving emails from it, send an email to reviewboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
RB 2.0.17 - Interdiff fails to highlight deleted code
Hi all, I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff viewer. Steps to reproduce: 1. Change a few files (I'm using SVN) 2. rbt post changelist 3. Publish review 4. Delete a group of lines from one of the files already changed. 5. rbt post -r ID changelist 6. Publish review 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: PASS 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - Result: PASS 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - Result: FAIL In the second interdiff, RB states that the file contains only whitespace changes, this is incorrect as lines have been deleted. Expanding the entire file, the contain on the left and right diff panels is correct, I can see that the deleted lines have gone, however they're not highlighted in red as expected. I don't think this is related to caching as the following steps had no effect, the second interdiff consistently shows the same result. * systemctl restart memcached * systemctl restart httpd * View second interdiff with different browsers (Firefox and Chromium), both show same results, hence not a browser cache issue. I'm running ReviewBoard on CentOS7 using EPEL packages. Let me know if you need more information to help recreate or fix this bug. Thanks, Paul -- Supercharge your Review Board with Power Pack: https://www.reviewboard.org/powerpack/ Want us to host Review Board for you? Check out RBCommons: https://rbcommons.com/ Happy user? Let us know! https://www.reviewboard.org/users/ --- You received this message because you are subscribed to the Google Groups reviewboard group. To unsubscribe from this group and stop receiving emails from it, send an email to reviewboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: RB 2.0.17 - Interdiff fails to highlight deleted code
Hi Paul, We recently fixed a bug like this, but perhaps there's another issue somewhere. In order to diagnose this, I'll need to have original copies of the affected files, as well as both diffs. I won't be able to diagnose without those, unfortunately. Christian -- Christian Hammond - christ...@beanbaginc.com Review Board - https://www.reviewboard.org Beanbag, Inc. - https://www.beanbaginc.com -Original Message- From: Paul Fee paul.f@gmail.com Reply: reviewboard@googlegroups.com reviewboard@googlegroups.com Date: June 30, 2015 at 4:02:10 AM To: reviewboard@googlegroups.com reviewboard@googlegroups.com Subject: RB 2.0.17 - Interdiff fails to highlight deleted code Hi all, I'm using ReviewBoard 2.0.17 and see unexpected behaviour in the diff viewer. Steps to reproduce: 1. Change a few files (I'm using SVN) 2. rbt post 3. Publish review 4. Delete a group of lines from one of the files already changed. 5. rbt post -r 6. Publish review 7. Review entire diff: http://reviewboard/r/9581/diff/2/ - Result: PASS 8. Review first interdiff: http://reviewboard/r/9581/diff/1/ - Result: PASS 9. Review second interdiff: http://reviewboard/r/9581/diff/1-2/ - Result: FAIL In the second interdiff, RB states that the file contains only whitespace changes, this is incorrect as lines have been deleted. Expanding the entire file, the contain on the left and right diff panels is correct, I can see that the deleted lines have gone, however they're not highlighted in red as expected. I don't think this is related to caching as the following steps had no effect, the second interdiff consistently shows the same result. * systemctl restart memcached * systemctl restart httpd * View second interdiff with different browsers (Firefox and Chromium), both show same results, hence not a browser cache issue. I'm running ReviewBoard on CentOS7 using EPEL packages. Let me know if you need more information to help recreate or fix this bug. Thanks, Paul -- Supercharge your Review Board with Power Pack: https://www.reviewboard.org/powerpack/ Want us to host Review Board for you? Check out RBCommons: https://rbcommons.com/ Happy user? Let us know! https://www.reviewboard.org/users/ --- You received this message because you are subscribed to the Google Groups reviewboard group. To unsubscribe from this group and stop receiving emails from it, send an email to reviewboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. -- Supercharge your Review Board with Power Pack: https://www.reviewboard.org/powerpack/ Want us to host Review Board for you? Check out RBCommons: https://rbcommons.com/ Happy user? Let us know! https://www.reviewboard.org/users/ --- You received this message because you are subscribed to the Google Groups reviewboard group. To unsubscribe from this group and stop receiving emails from it, send an email to reviewboard+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.