Mailing lists / fpc-devel / 2017-06
32 messages · 7 people · –
Reply by email Subscribe to this list Your subscription
Read-only archive. Use j and k to move between messages.
While I have not done much the last month, I did keep up with most of the merging so most requests have been honoured, and I plan to branch mid next week or even monday if there is not much response.
If I missed some, or there are important new ones, there is still a few days time, waiting for the win32/64 findfirst fix.
Note that compiler/language merges need approval and an estimate of mergability from a compiler devel.
Op 17-06-17 om 21:31 schreef Marco van de Voort:
I'm hesitating about merging all fpmkunit and fppkg changes. Things improved a lot, but a little bit more testing could be useful. Not that it's used a lot, though.
At least it should handle https, since idefix is https only, now.
What do you think?
Joost.
While I have not done much the last month, I did keep up with most of the merging so most…
Another merge request:
r36498 - the fix for the fpwidestring bug, which caused the #0 in the compiler output of fpc-trunk. Even though the fpc-fixes compiler doesn't output #0 characters (because it doesn't have code page aware compiler messages), the actual bug is in fpwidestring's conversion from widestring to utf-8 and it still exists in fpc-fixes, therefore this merge request... :)
Nikolay
While I have not done much the last month, I did keep up with most of the merging so most…
Below a really long list. They are patches that happened after tag of 3.0.2, so they suppose that all earlier patches to fcl-pdf, fcl-js, pas2js, fcl-passrc have been merged already (which, I think, should be the case).
Output from svn log, so I assume it is somehow scriptable.
Michael.
Everything related fcl-pdf:
r35519
r36354
r36355
r36356
r36357
r36358
Everything related to fcl-js:
r35402
r35416
r35428
r35471
r35490
r35502
r35576
r35577
r35633
r35640
r35651
r35789
r36235
r36471
r36492
r36493
r36494
Everything related to fcl-passrc:
r35391
r35402
r35415
r35428
r35468
r35469
r35470
r35487
r35488
r35489
r35503
r35522
r35524
r35562
r35571
r35578
r35579
r35580
r35581
r35582
r35584
r35586
r35588
r35591
r35593
r35597
r35612
r35613
r35614
r35616
r35617
r35621
r35625
r35631
r35633
r35635
r35636
r35637
r35638
r35639
r35641
r35642
r35648
r35652
r35667
r35680
r35681
r35682
r35691
r35692
r35694
r35695
r35697
r35702
r35703
r35708
r35709
r35710
r35713
r35715
r35718
r35719
r35720
r35728
r35731
r35735
r35737
r35782
r35790
r35791
r35793
r35794
r35798
r35801
r35806
r35808
r35810
r35827
r35829
r35839
r35842
r35844
r35846
r35851
r35853
r35859
r35863
r35865
r35866
r35868
r35870
r35872
r35874
r35876
r35883
r35884
r35887
r35889
r35895
r35896
r35897
r35898
r35902
r35904
r35909
r35910
r35912
r35914
r35915
r35916
r35917
r35923
r35926
r35928
r35931
r35933
r35938
r35940
r35947
r35948
r35949
r35950
r35951
r35960
r35963
r35964
r35966
r36006
r36069
r36084
r36117
r36121
r36125
r36126
r36127
r36129
r36130
r36131
r36140
r36154
r36155
r36160
r36161
r36171
r36177
r36178
r36179
r36181
r36182
r36183
r36184
r36185
r36186
r36187
r36188
r36189
r36190
r36192
r36205
r36213
r36216
r36217
r36218
r36219
r36220
r36221
r36222
r36223
r36224
r36225
r36226
r36228
r36234
r36237
r36238
r36241
r36246
r36248
r36318
r36320
r36376
r36397
r36453
r36458
pas2js related:
r35402
r35417
r35428
r35472
r35491
r35504
r35574
r35583
r35585
r35587
r35615
r35631
r35633
r35643
r35653
r35668
r35683
r35693
r35696
r35697
r35704
r35705
r35706
r35711
r35714
r35716
r35718
r35719
r35729
r35732
r35736
r35738
r35795
r35799
r35802
r35803
r35804
r35805
r35809
r35811
r35828
r35840
r35843
r35845
r35847
r35852
r35854
r35860
r35862
r35864
r35867
r35869
r35871
r35873
r35875
r35888
r35903
r35905
r35908
r35911
r35917
r35919
r35920
r35927
r35932
r35939
r36007
r36035
r36037
r36070
r36085
r36118
r36154
r36156
r36172
r36236
r36242
r36247
r36319
r36459
r36460
While I have not done much the last month, I did keep up with most of the merging so most…
This is now fixed, rev. 36510.
I will still check the svn log of rtl/packages for any fixes that may be merged.
Unless you have somewhere an up-to-date list of possible candidates ?
Michael.
Below a really long list. They are patches that happened after tag of 3.0.2, so they…
Op 18-06-17 om 09:34 schreef Michael Van Canneyt:
Which reminds me...
We use fcl-passrc to use pascalscript to validate form-input on websites. We evaluate the scripts on the server-side, but also direct on the client. With a very simple pascalscript to js converter.
But it doesn't work with trunk anymore, at least, it doesn't parse the statement 'false;', together with 'true;' anymore. Which is used a lot in our system. (For example for the 'required'-statements.)
Is this intentional?
Regards,
Joost.
fpc gives "Illegal expression".
Maybe pascalscript needs a flag in pparser to allow this.
Mattias
This is now fixed, rev. 36510. I will still check the svn log of rtl/packages for any…
In our previous episode, Michael Van Canneyt said:
The usual weblocation? http://www.stack.nl/~marcov/mergelogs26
Some quick checks (greps for the searchterm in commitlog+ filenames)
fcl-pdf eligible: 33713,34002,34008,34615,35278,35833,36395
fcl-passrc eligible:
29399,29950,29974,30891,31177,31318,31527,31857,34002,34008,35833,36006,36020,36069,36084,36117,36121,36125,36126,36127,36129,36130,36131,36140,36154,36155,36160,36161,36171,36177,36178,36179,36181,36182,36183,36184,36185,36186,36187,36188,36189,36190,36192,36205,36213,36216,36217,36218,36219,36220,36221,36222,36223,36224,36225,36226,36228,36234,36237,36238,36241,36246,36248,36318,36320,36321,36376,36395,36397,36453,36458
pas2js:
29399,29950,29974,30891,31177,31318,31525,31527,34002,34008,35833,36035,36037,36085,36118,36154,36156,36172,36236,36242,36247,36319,36391,36459,36460
etc etc.
The 'merge sets' are followed by a pair of numbers. What is the meaning of these numbers ?
Then if I open 'fcl-pdf', I see:
(inactive) Revisions in this set 33162,33163,33175,33176,33212,33215,33255,33264,33401,33428,33431,33453,33456,33468,33482,33483,33484,33485,33495,33497,33499,33535,33543,33563,33564,33566,33573,33998,34037,34060,34543,34563,34767,34778,34779,34780,34781,34804,35083,35084,35090,35094,35126,35131,35519
What does (inactive) mean ?
Concerning the eligble below, everything can be merged.
fcl-image: everything can be merged.
fcl-web: everything can be merged.
fpdoc: everything can be merged.
I will check out fcl-base.
Michael.
In our previous episode, Michael Van Canneyt said:
> Some quick checks (greps for the searchterm in commitlog+ filenames) The 'merge sets' are followed by a pair of numbers. What is the meaning of these numbers ?
Active/inactive
Then if I open 'fcl-pdf', I see: (inactive) Revisions in this set 33162,33163,33175,33176,33212,33215,33255,33264,33401,33428,33431,33453,33456,33468,33482,33483,33484,33485,33495,33497,33499,33535,33543,33563,33564,33566,33573,33998,34037,34060,34543,34563,34767,34778,34779,34780,34781,34804,35083,35084,35090,35094,35126,35131,35519 What does (inactive) mean ?
Added to the set, but disappeared from eligible after that. In other words: already merged. Only active ones matter.
Okay, thanks.
You may want to add this info as a 'legend' to the pages, because it is not readily obvious (well, not to me anyway). Maybe also put the inactive ones at the end...
Michael.
In our previous episode, Michael Van Canneyt said:
You may want to add this info as a 'legend' to the pages, because it is not readily obvious (well, not to me anyway). Maybe also put the inactive ones at the end...
I added a line, but the most important thing to understand is that the revision sets are revisions that I "isolated" from the "merged not in other sets" category.
So all new revs enter the "to be merged not in other sets" category, and when I do queries I get lists of revs that go into the query categories.
In our previous episode, Marco van de Voort said:
So all new revs enter the "to be merged not in other sets" category, and when I do queries I get lists of revs that go into the query categories.
.... so e.g. newer fcl-base revs might still be in the to be merged category. The categorization is all manual. (since if you do it in an automated way, revs that touch many directories like makefile or fpmake related end up in random categories)
While I have not done much the last month, I did keep up with most of the merging so most…
Can you merge r35878 as well please?
It is a small fix for TAssert.AssertEquals with strings.
Thanks,
Denis
Op 17-06-17 om 21:31 schreef Marco van de Voort: I'm hesitating about merging all…
In our previous episode, Joost van der Sluis said:
We could have take the risk just after a release, so we had some months for problems to emerge. Last minute merge of 118 revs that are not regularly tested in trunk is a bit minor. Moreover fpmkunit is heavily used due to fpmake, and merging is a pain, since support for newer targets that don't exist in fixes might need manual adjusting.
So I would guess try to name hotfixes only.
That is a bit unfair, since I do merge fcl-passrc atm, but that package is currently quite used heavily in trunk, and I kept up with merging every 2 months or so without reported problems.
Can you merge r35878 as well please? It is a small fix for TAssert.AssertEquals with…
In our previous episode, Denis Kozlov said:
Can you merge r35878 as well please? It is a small fix for TAssert.AssertEquals with strings.
Done.
In our previous episode, Joost van der Sluis said: We could have take the risk just after…
Op 19-06-17 om 10:56 schreef Marco van de Voort:
No probem, I can live with that. It's clear this way.
Also, the changes are only stable for about a few weeks. Merging it earlier was not really an option.
I'll see if I can make a list with hotfixes. Which will be a pain to merge, as the code in which the fixes are done might even not exist anymore.
But I think we have to change http into https in a few places...
Regards,
Joost.
Op 19-06-17 om 22:52 schreef Joost van der Sluis:
Ok, please merge:
32732
32759
33016
33017
33033 *
34141
34142
34144
34145
34190
34314
34396
34401 *
34403 *
34486
35277 (If difficult to merge, skip)
35881
36511
36529 *
36550 (brand new, but needed to handle https downloads and redirects) 36551 (idem)
36552 (idem)
36553 (idem)
Marked with * are really important. The last few are just small changes, but necessary for the https-change. It could be that they don't merge due to earlier changes. In that case, just change 'http' to 'https' and 'lnet' to 'FPC' manually....
Regards,
Joost.
In our previous episode, Joost van der Sluis said: [ Charset UTF-8 unsupported, converting... ]
The first four merge, but the fifth has conflicts that are easily resolved but won't compile. r32022 ?
I did a query for fpmkunit.pp and added it to the mergelogs fppkg /fpmkunit category, so that all fpmkunit.pp revs are 100% there.
In our previous episode, Marco van de Voort said:
.. which has tie-ins to unmerged TProcess revisions.
Then skip them. As you can see they solve TProcess-related problems on Windows. Maybe these problems are not there in fixes.
Or.. we have serious problems with TProcess on Windows, and the fix therefor is not fixed yet. Maybe that Pierre can help with that?
Regards,
Joost.
In our previous episode, Joost van der Sluis said:
They are new features, not fixes:
* patch by Mattias Gaertner to make TProcess more unicode aware, resolves issue #29136
* Add poPassInput to be able to handle interactive processes
* improved TProcess.WaitOnExit with timeout implementation to use fpGetTimeOfDay to work more precise
* Do not call CloseHandle on hStdInput if poPassInput is in Options
While I have not done much the last month, I did keep up with most of the merging so most…
ptc merge request:
r34598, r35481, r36507, r36508, r36509, r36555, r36556
Nikolay
In our previous episode, Nikolay Nikolov said:
Merged. The first is the FSF address update, so I got initially a lot of files :-)
Okay, thanks. You may want to add this info as a 'legend' to the pages, because it is not…
For fcl-base, I think
34498,34499,34506,34509, 35518,35742
can be merged.
Michael.
Okay, thanks. You may want to add this info as a 'legend' to the pages, because it is not…
Reviewed some more lists:
fcl-db:
34464,34468,34469,34510,34962,34963,34980,35209,35419,35486,35899
fcl-web:
35297,25563,35821
non-categorized:
31464,31467,31468,
32271,32326,32327,32332,32369,
33210,33363,33695,
34097,34238,34239,34240,34245,34475,34476,34481,34483,34484,34487,34489,34490,34493,34494,34495, 34501,34502,34505,34507,34511,34518,34683,34725,34817,34834,34847,34866,34868,34873,34874,34884, 34903,34905,34925,34968,34966,34968,34969,34970,34983,34986,34997, 35063,35331,35332,35333,35342,35356,35398,35422,35423,35467,35474,35517,35560,35598, 35601,35603,35611,35619,35620,35622,35624,35892,35901,35913,35946,35954, 36062,36214,36215,36274,36406
Michael.
In our previous episode, Michael Van Canneyt said:
>> already merged. Only active ones matter. Reviewed some more lists:
It was last call for emergency revs the day before release branching, not everything you could find :-)
Maybe I should have made that clearer. The changes of Nikolay e.g. are only local to one package that isn't used that much.
The db changes have been vetted by lacak already and
fcl-web: 35297,25563,35821
e.g.
35821 is blocked uses 3.1.1 language features. (due to bug #32047 in the ini handling of 3.0.2+ the warnings don't show up anymore it seems)
the others are about new packages that are then enabled for all OS the day before branchpoint? Not really a good idea.
Nothing urgent.
Also contains patches already rejected for 3.0.2 like the first ones.
There might be some cherry picking doable here, but is that really worth delaying the release?
Reviewed some more lists: fcl-db: 34464,34468,34469,34510,34962,34963,34980,35209,35419,35…
Reviewed some more lists:
fcl-db: 34464,34468,34469,34510,34962,34963,34980,35209,35419,35486,35899
34962,34963 - this was already merget as a special patch IIRC 35209 - IMO depends on changes in sqldb (which are related to code page awara TStringField, TMemoField)
L.
In our previous episode, Michael Van Canneyt said: It was last call for emergency revs…
No problem, leave it as-is then :)
Michael.
In our previous episode, Michael Van Canneyt said:
I branch at home tonight. I'll see if I can do some of the more harmless fixes of the last list (like the comment only changes, the openssl changes)
Great, many thanks !
Michael.
Another merge request: r36498 - the fix for the fpwidestring bug, which caused the #0 in…
Are you planning to merge this?
Nikolay