https://pfspipe.ipmu.jp/jira This file is an XML representation of an issue en-us 8.3.4 803005 13-09-2019 https://pfspipe.ipmu.jp/jira/browse/INSTRM-1415 Instrument control development <p>We want the SPS, FPS/MCS, and AGC visits to be synchronized by PFI convergence: for the visit used for the `fps moveToPfsDesign` command to be the one which starts that field's SPS and AGCC visits.</p> INSTRM-1415

Add visit0 logic to IIC

Task Normal Done Done arnaud.lefur cloomis EngRun Fri, 22 Oct 2021 21:12:07 +0000 Thu, 13 Jan 2022 18:32:53 +0000 Thu, 13 Jan 2022 18:32:53 +0000 ics_iicActor 0 4 <p>This logic has also to take into account that contrary to fps, other lightSource config change(dcb bundles, sunss pointing) is not necessarily associated with a visit. <br/> it gets a visit0 when someone ask for a sps exposure basically.</p> <p>okay, so that's 90% done, but I'm hitting an opdb issue that cannot be worked around.<br/> basically between <em>iic_sequence</em> and <em>visit_set</em> table.<br/> <em>iic_sequence</em> has a unique <em>visit_set_id</em> and <em>visit_set</em> has <em>pfs_visit_id</em> as primary key which means that a given <em>pfs_visit_id</em> can <b>only</b> be associated with one <em>visit_set_id</em>, which made sense with independent visits.</p> <p>But since we're sharing <em>pfs_visit_id</em> among fps,sps,agc, and have a <em>visit_set_id</em>, for one of those, you want to insert several sequence for a given visit which break the constraint.</p> <p>One solution is to </p> <ul> <li>add a <em>visit0</em> column to <em>iic_sequence</em></li> <li>change <em>visit_set_id</em> to <em>visit0</em> in <em>visit_set</em> (without back-populating).</li> </ul> <p>This way, you keep the unicity of <em>pfs_visit_id</em>,<em>visit0</em> in <em>visit_set</em> but different <em>iic_sequence</em> can share the same <em>visit0</em></p> <p>In this case, renaming <em>visit_set_id</em> to <em>sequence_id</em> would probably make more sense, but do we care ? I'm not sure.</p> <p>okay last proposal after discussions with <a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=cloomis" class="user-hover" rel="cloomis">cloomis</a> and <a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=rhl" class="user-hover" rel="rhl">rhl</a>.</p> <ul> <li>rename iic_sequence.visit_set_id to iic_sequence.sequence_id</li> <li>add iic_visit0 table with iic_sequence.sequence_id as primary key, and visit0</li> <li>rename visit_set.visit_set_id to visit_set.visit0</li> </ul> <p>I'm pulling opdb table from the summit and preparing a script to do those schema changes which I will hand off to <a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=kiyoto.yabe" class="user-hover" rel="kiyoto.yabe">Kiyoto Yabe</a> when I'm sure it's doing the right thing.</p> <p>That seems to be probably reasonable to me.</p> <p>`sequence_id` is totally different from `status_sequence_id` right? Can we make it as `iic_sequence_id` to avoid any confusion? </p> <p>A script for the schema change would be very helpful, but I prefer to make the actual change through the alembic mechanism after fixing `models.py`.</p> <p>This change probably affects the `obslog` so I took the liberty of inviting <a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=michitaro" class="user-hover" rel="michitaro">michitaro</a> for further discussion.</p> <p>in the end, that scheme was again not quite right, we're confusing the official visit0 with the first visit of the visit_set which is sometimes equivalent but sometimes not.</p> <p>the final proposal is actually much less disruptive than the previous one which is always good.</p> <p>we're adding a <em>field_set</em> table :</p> <ul> <li>visit_set_id = Column(Integer, ForeignKey('iic_sequence.visit_set_id'), primary_key=True)</li> <li>visit0 = Column(Integer, ForeignKey('pfs_config.visit0'))</li> </ul> <p>I also add to add a unique constraint on pfs_config.visit0 otherwise it's complaining the foreign key is not unique :</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>ProgrammingError: (psycopg2.errors.InvalidForeignKey) there is no unique constraint matching given keys for referenced table "pfs_config" </pre> </div></div> <p>I check a bit more, it's true that currently there is no explicit unique contraint :</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre> CONSTRAINT pfs_config_pkey PRIMARY KEY (pfs_design_id, visit0), CONSTRAINT pfs_config_pfs_design_id_fkey FOREIGN KEY (pfs_design_id) REFERENCES public.pfs_design (pfs_design_id) MATCH SIMPLE ON UPDATE NO ACTION ON DELETE NO ACTION </pre> </div></div> <p>compared to </p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre> CONSTRAINT pfs_config_pkey PRIMARY KEY (pfs_design_id, visit0), CONSTRAINT pfs_config_pfs_design_id_fkey FOREIGN KEY (pfs_design_id) REFERENCES public.pfs_design (pfs_design_id) MATCH SIMPLE ON UPDATE NO ACTION ON DELETE NO ACTION, CONSTRAINT pfs_config_visit0_key UNIQUE (visit0) </pre> </div></div> <p>I'm a bit confused about using _<em>table_args</em>_ to pass contraints, it's look like it's not completely equivalent to declare a column unique.<br/> <a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=kiyoto.yabe" class="user-hover" rel="kiyoto.yabe">Kiyoto Yabe</a>, you have a better expertise than I do about sqlAlchemy, I will ask you to review this ticket.</p> <p>Changes look fine to me.</p> <p>Yabe-san has approved the changes.</p> Blocks INSTRM-1422 Development Rank 0|02qpqj:m Reviewers kiyoto.yabe Sprint PreEngRun4 Story Points 2.0